Everything You Need to Know About Training Volume

Several months ago, I published an article titled “Strength Data Don’t Tell You Much About Hypertrophy.”

I wrote that article for two reasons.

First, on the most basic level, I simply wanted to push back against a common argument that’s been cropping up more and more frequently: strength data tells you more about hypertrophy than hypertrophy data does.

Second, I wanted to see if it would prompt proponents of that position to reflect on and refine their stance, in order to present strong counterarguments or data to support their position. I knew I would be writing this article as a follow-up, so I wanted to be sure I wasn’t overlooking any relevant data, or misunderstanding any nuances of that position.

This ongoing discussion has primarily revolved around the topic of training volume. Namely, a contingent of people would like to argue that muscle growth is maximized at relatively low training volumes. The research directly investigating the impact of training volume on muscle growth does not support this position whatsoever. However, you can make the case that strength gains are maximized at relatively low training volumes. So, low-volume proponents contend the strength data implies that muscle growth must also be maximized at relatively low training volumes, and any apparent increases in muscle growth at higher volumes are merely the result of muscle swelling. I believe that position is incorrect, and that we can thoroughly demonstrate why it’s incorrect.

I’d like to briefly discuss what I’d like to accomplish with this article. On the most basic level, this article is going to discuss the impact of training volume on muscle growth and strength gains. Going one level deeper, this article will aim to refute the main arguments of the folks who contend that muscle growth is maximized with relatively low training volumes, and discuss why we can be quite confident that higher volumes do actually increase muscle growth. Finally, this article will (hopefully) help you think a bit more critically about research interpretation and scientific epistemology.

Before we get rolling, I also want to make sure we’re all on the same page about some terminology. Throughout this article, when I talk about training volume, unless it’s otherwise specified, I’m referring to weekly fractional sets. So, we’re focusing on weekly training volume, not per-session training volume. To determine “fractional sets,” each set of an exercise counts as one set of volume for muscles that are primarily targeted by that particular exercise, and half a set of volume for muscles that are “meaningfully trained but not the primary force generator of the exercise (i.e., synergist).” So, for example, each set of biceps curls would count as one set of weekly volume for the biceps, but each set of rows would only count as half a set of weekly volume for the biceps. Furthermore, unless otherwise stated, you can assume that the sets being discussed are sets that are taken all the way to the point of concentric failure (or very close to it). Finally, to provide some rough volume landmarks: the “low volume” position is that hypertrophy is maximized with 10 or fewer sets per week (the most common number you’ll see these days is 8 sets, and the argument employed by these folks to justify their belief would imply that hypertrophy should be maximized with just 5 sets, but for the purposes of this article, ≤10 sets will serve as our low-volume catch-all); the middle-of-the-road opinion – and probably still the most common opinion – is that hypertrophy is maximized with 10-20 sets per week; and the high-volume position is that hypertrophy is maximized with >20 sets per week.

So, with that out of the way, let’s dive in!

The typical understanding of the relationship between muscle growth and strength gains

The predominant understanding of the impact of training volume on both muscle growth and strength gains has been most heavily informed by a recent series of meta-regressions by Pelland and colleagues.

For a study to be included in these meta-regressions, it needed to meet the following criteria:

1. The study needed to compare the effects of at least two different levels of training volume, with other relevant training variables (proximity to failure and relative load) held constant between groups or conditions.
2. The study needed to include a longitudinal training intervention, using resistance exercise including both an eccentric and concentric component, and lasting for at least four weeks.
3. The subjects needed to be healthy and non-elderly (<70 years old).
4. The study needed to assess changes in muscle size using site-specific measures (i.e., ultrasound, MRI, or muscle biopsy, rather than just assessing fat-free mass), and/or assess changes in maximal dynamic (1-10RM), isokinetic, or isometric strength.

Ultimately, 35 studies met these inclusion criteria for hypertrophy outcomes, and 66 studies met these inclusion criteria for strength outcomes.

For hypertrophy, they found that higher volumes tended to lead to more muscle growth.

For strength, on the other hand, they found that higher training volumes only led to larger strength gains to a point.

Past about 5 sets per week, additional strength gains with further increases in volume were smaller than the “smallest detectable effect size.”

Simple enough, right? As training volumes increase, muscle growth increases, albeit with diminishing returns. However, as training volume increases, strength gains only increase up to about 4 or 5 sets per week. Past that point, if further increases in volume lead to larger strength gains, the marginal benefit is too small to reliably detect.

Why this understanding of the volume literature has led to consternation

These findings understandably led to some confusion. After all, muscle is contractile tissue. If you build more contractile tissue, your muscles should be able to generate more force. If your muscles can generate more force, that increase in force output should manifest as a larger increase in strength measurements, right?

As discussed in the first article in this series, that’s not how it always works. Muscle mass and strength have a surprisingly complex relationship, and there’s surprisingly little direct evidence supporting the idea that muscle growth causally increases strength gains (though, I certainly think it does).

Furthermore, as I’ll discuss in a moment, there’s a pretty obvious reason why you shouldn’t necessarily expect the relationships in those two meta-regression models to be transitive or additive. In other words, they tell you the general relationship between volume and strength gains in isolation, and they tell you the general relationship between volume and hypertrophy in isolation, but they don’t tell you much about the joint relationship between hypertrophy and strength gains at different levels of volume (or the impact – or lack of impact – of hypertrophy on strength gains at different levels of volume).

But, before I get too far ahead of myself, I first need to briefly explain the strength results people would have expected to see, if higher training volumes truly lead to more muscle growth.

Strength is often conceptualized as the simple product of two factors: muscle size and motor skill (or “neural adaptations”). When comparing strength levels between individuals, there are more factors that come into play (leverages, muscle insertions, etc.), but it’s typically assumed that those other factors can be ignored when assessing strength changes in a single lifter over time.

The assumption that strength is the product of muscle size and motor skill leads to a pair of predictions:

1. Gains in muscle size should directly contribute to strength gains on a proportional basis. In other words, if your muscles get 10% larger, you should expect to get 10% stronger.
2. In situations where you’d expect “neural adaptations” to occur, or situations where you’d expect general motor skill with a particular exercise to increase, these adaptations should result in additional strength gains, such that total gains in strength exceed gains in muscle size.

In other words, if you’re doing an exercise you’re already quite skilled at, increasing your muscle mass by x% should lead to an x% increase in strength. However, if you’re doing an exercise you’re not yet particularly skilled at, such that muscle mass still increases by x%, and “neural adaptations” occur that independently increase your strength performance in that exercise by y%, your total strength gains should be x% + y%. In essence, relative gains in strength should almost always either equal or exceed relative gains in muscle size for exercises you perform regularly in your training.

So, when gains in muscle size exceed gains in strength, this divergence therefore warrants an explanation. Given the set of assumptions we’re working with here, there only appear to be two potential explanations:

1. Motor skill or “neural adaptations” for the exercise used to assess strength somehow regressed (I’d lump accumulated fatigue in here as well).
2. The observed increase in muscle size doesn’t actually reflect a proportional increase in contractile protein content.

In most research contexts, option 1 is unlikely, so the natural impulse is to focus on option 2.

A tacit assumption underpinning the belief that gains in strength should scale proportionately to gains in muscle size is the assumption that the relative contractile protein content of a muscle remains steady as the muscle grows. Muscle size, in this case, is a proxy for the number of parallel contractile units (myofibrils, composed of actin and myosin proteins) within the muscle. If contractile protein density remains consistent, you should expect proportionate strength gains. If contractile protein density increases (more contractile proteins per unit of muscle cross-sectional area), you’d expect gains in strength to exceed gains in size. However, if contractile protein density decreases (fewer contractile proteins per unit of muscle cross-sectional area), you’d expect gains in strength to lag behind gains in size.

So, when gains in muscle size exceed gains in strength, it’s often assumed that the contractile protein density of the muscle tissue must have decreased as well. This decrease in contractile protein density may be the result of sarcoplasmic hypertrophy (muscle fiber growth where the expansion of non-contractile elements outpaces the expansion of the contractile protein fraction of the fiber), or may reflect muscle swelling resulting from muscle damage and the resulting inflammatory response (leading to a temporary increase in fluid content within the muscle), but either way, many people assume that the observed increase in muscle size can’t possibly be fully accounted for by “real” myofibrillar hypertrophy.

So, circling back to the Pelland meta-regressions, it appears that further increases in training volume past 5 sets per week fail to lead to further increases in strength. Given the assumptions discussed in this section, this plateau in strength gains is thought to suggest that gains in “real” myofibrillar hypertrophy also plateau at relatively low volumes. However, the hypertrophy meta-regression suggests that higher volumes continue to lead to more growth. So, it’s therefore assumed that the additional “growth” observed with higher training volumes reflects gains in non-contractile elements in the muscle: sarcoplasmic hypertrophy or (more often) muscle swelling and edema.

Formally, here’s how the arguments against the hypertrophic benefits of higher training volumes are constructed. I’ll discuss them in a more formal way later (discussing the structure of the arguments, specific terminology, etc.). But for now, it’ll be good to see them laid out clearly and succinctly:

Argument 1:

Premise 1: Muscular strength capacity is fully determined by the number of parallel actin/myosin crossbridges aligned with the muscle’s line of pull.

Premise 2: Myofibrillar hypertrophy is characterized by an increase in the total number of parallel actin/myosin crossbridges in a muscle.

Conclusion 1: Therefore, myofibrillar hypertrophy increases muscular strength capacity, and an increase in muscle size that’s not accompanied by an increase in strength capacity cannot reflect myofibrillar hypertrophy.

Argument 2:

Premise 1: Myofibrillar hypertrophy increases muscular strength capacity, and an increase in muscle size that’s not accompanied by an increase in strength capacity cannot reflect myofibrillar hypertrophy (the conclusion of Argument 1 is a premise in Argument 2).

Premise 2: Higher training volumes fail to cause larger strength gains when volumes exceed ~5 sets per week.

Conclusion 2: Therefore, higher training volumes (past ~5-10 sets per week) cannot actually be promoting greater myofibrillar hypertrophy.

Inference flowing from Conclusion 2: The apparent increases in hypertrophy at higher training volumes must therefore be the result of muscle swelling, edema, or sarcoplasmic hypertrophy.

Apologies if this section was almost insultingly basic for many readers. I thought it was necessary to include for two reasons. First, there will be people who read this article for whom this is all new information. Second, I wanted to demonstrate to the people who will inevitably disagree with my conclusions in this article that I do understand their arguments, and I’m not just trying to take down a strawman.

So now, let’s shift gears. I’ve explained why many people believe that these meta-regressions suggest that muscle growth can be maximized at relatively low training volumes (and, as a corollary, why they believe that the additional “hypertrophy” observed with higher training volumes primarily reflects increases in non-contractile elements in the muscle). I wouldn’t even be surprised if many readers find the arguments in this section to be logical and persuasive. So, now it’s time to explain why the data don’t actually support the conclusions people are trying to draw, followed by pointing out a fairly large hole in this line of thinking.

Higher training volumes do actually lead to larger strength gains

Apples and oranges

As I previously alluded to, there’s a pretty obvious reason why you shouldn’t make direct comparisons between the strength and hypertrophy meta-regressions in the Pelland paper. The reason: the hypertrophy meta-regression and strength meta-regression are based on different sets of studies.

The strength meta-regression included nearly twice as many studies with nearly twice as many subjects as the hypertrophy meta-regression. The strength meta-regression included 66 studies with a total of 2,020 participants, while the hypertrophy meta-regression included 35 studies with 1,032 participants. Furthermore, the two meta-regressions had just 21 studies in common. So, only 60% of the studies in the hypertrophy meta-regression were also used in the strength meta-regression, and only 32% of the studies in the strength meta-regression were also used in the hypertrophy meta-regression. There’s some overlap, but these analyses weren’t intended to show you the impact of volume on strength gains and muscle growth in a common set of studies.

Venn diagram illustrating the overlap of strength vs. hypertrophy studies included in the Pelland meta-regressions

In other words, if we see that higher training volumes lead to more muscle growth in one set of studies, and we also see that strength gains plateau at 5 sets per week in a (mostly) different set of studies, we can’t necessarily conclude that training volume (and the hypertrophic effects of higher training volume) stops contributing to strength gains at 5 sets per week. To make that conclusion, we’d need to see that higher volumes didn’t lead to larger strength gains in the studies finding that higher volumes led to more muscle growth.

Notably, many of the non-overlapping studies were studies in untrained subjects (mostly) finding large strength gains while (mostly) testing fairly low training volumes.

Here’s a simple illustration of how that could present a problem.

Let’s just assume that both strength and hypertrophy increase linearly as training volumes increase (to be clear, I don’t believe that – this is just to make the illustration easier to understand). You have a set of 20 studies in trained lifters that assess both muscle growth and strength gains in the same pool of subjects. When you analyze these 20 studies, it’s clear that gains in muscle size run parallel with gains in strength.

However, what might happen if you then added an additional batch of 20 studies that only assess changes in strength in groups of untrained lifters? These studies use fairly low training volumes, but the subjects still gain a lot of strength because, as untrained lifters, they have a lot of strength to gain purely as a result of “neural adaptations.” Furthermore, since “neural adaptations” are the predominant contributor to strength gains in these subjects, gains in strength are less impacted by training volume or muscle growth, and are more heavily impacted by the complexity of the exercises used to assess strength. In studies that assess strength gains in relatively simple exercises like biceps curls, relative strength gains are smaller since there’s less of a learning curve, but in studies that assess strength gains in more complex exercises like squats, relative strength gains are larger since the learning curve is much steeper. As a result, the strength data from these studies is quite a bit noisier.

With the addition of these 20 non-overlapping “studies,” it may suddenly appear as if the strength benefits of increasing your training volume begin to plateau at much lower levels of volume.

So, to mitigate this possibility, I downloaded the dataset from the Pelland meta-analysis, and checked the relationship between training volume and hypertrophy, and the relationship between training volume and strength gains in studies meeting two simple criteria:

1. The studies needed to use trained subjects.
2. The studies needed to include both hypertrophy and strength measures that are directly related to each other (for example, measuring both squat 1RM and quadriceps thickness, or bench press 1RM and triceps thickness) in the same group of subjects.

The reason for these criteria are pretty simple.

First, in studies on untrained subjects, we know that there’s very little relationship between strength gains and muscle growth, because most of the strength gains people experience during the first few months of lifting simply come from gaining proficiency with the movement patterns used to assess strength. By focusing on trained subjects, we’re focusing on the population where we’d most expect hypertrophy to have a direct, measurable influence on strength gains.

Second, only focusing on studies that include related assessments of both hypertrophy and strength gains allows us to make a more direct “apples to apples” comparison between the impact of training volume on muscle growth, and the impact of training volume on strength gains. With a shared study pool for both analyses, you can be a lot more confident in any inferences you make about the relationship between muscle growth and strength gains at varying levels of training volume. You’re comparing hypertrophy results to strength results in the same groups of subjects, and in the specific strength assessments that should be directly impacted by muscle growth.

Below, you can find a list of the studies and measurements meeting these criteria. Ultimately, we’re dealing with a pretty decent sample size: 35 paired measures of strength and hypertrophy within 22 unique groups of subjects across 8 studies.

When we look at the volume literature through this lens (which is, I think, the correct lens if we’re interested in determining whether the increase in muscle size resulting from higher training volumes impacts the strength gains achieved when training with higher volumes), a considerably different picture emerges: higher training volumes are similarly associated with more muscle growth and larger strength gains (r = 0.52 for strength gains, and r = 0.55 for muscle growth):

Of note, if seeing linear trendlines in this context makes you uncomfortable (to be clear, I don’t think that muscle growth and strength gains linearly increase as a function of volume), below you can see how the data looks when fitted with logarithmic trendlines instead. For what it’s worth, linear trendlines fit the data better (higher r² values), but I do still suspect that diminishing returns kick in as volume increases. Personally, I’m not too concerned about the exact style of the trendline that’s fit to the data – I’m much more interested in the high-level point (i.e., that strength gains in trained lifters don’t plateau at just 5 sets per week in studies where related measures of muscle growth also increase with higher training volumes).

Next, let’s analyze direct comparisons of muscle growth versus strength gains in each set of paired measurements in each group, in each study. In the regression analyses above, it appears strength gains exceed muscle growth on a relative basis at all levels of volume. Is that also what we observe when focusing on each pair of associated strength and hypertrophy measurements?

To see, I made a Bland-Altman plot to analyze the differences between paired measures of hypertrophy and strength gains at different levels of training volume. Each x-value is the average volume for each set of paired measures, and each y-value is the relative increase in strength, minus the relative increase in muscle size.

Here’s a simple example to illustrate how the following chart was made:

In a study by Brigatto and colleagues, the low-volume group did 12 sets per week contributing to triceps hypertrophy, and 16 sets per week contributing to gains in bench press 1RM. So, the average volume for these two measurements is 14 sets – that’s our x-coordinate. In this group, triceps thickness increased by 0.88%, while bench press 1RM increased by 23.65%. So, the difference between relative gains in strength and relative gains in muscle size was 23.65 – 0.88 = 22.77% – that’s our y-coordinate. So, we’d plot the point (14, 22.77), and move on to the next paired set of strength and hypertrophy measurements in the dataset.

If increases in volume have a larger impact on muscle growth than strength gains, you should expect the resulting graph to have an overall negative trend (like in the example below): at low volumes, there’s a large difference between gains in strength and gains in size, while at higher volumes, the gap is smaller, or the direction flips (i.e., gains in muscle size exceed gains in strength).

So, is that what we see in the actual data?

Not even close, as you can see in the graph below. The gap between gains in strength and gains in muscle size remains pretty consistent at all levels of training volume. If anything, there’s a slight positive slope (i.e., as volumes increase, gains in strength exceed gains in size to a larger and larger extent), but I wouldn’t pay much attention to that, since the association is quite weak (r = 0.23).

Of course, interesting patterns in individual studies may still be hiding within the aggregate data. So, let’s take a moment to examine the hypertrophy and strength gains observed in all of the studies with paired strength and hypertrophy measurements, where at least one group of subjects trained with quite high volumes for strength (at least 20 fractional sets per week). Within this group of studies, we can compare the strength gains and hypertrophy observed in the highest-volume groups to the second highest-volume groups within the same study. So, when we focus on these within-study comparisons, do we see that strength gains plateau at high volumes, while hypertrophy appears to continue increasing?

Not even close.

Overall, we have 9 strength measurements across 5 studies where at least one group of subjects trained with quite high volumes (at least 20 sets per week). Compared to groups of subjects in the same studies training with slightly lower volumes, the folks training with the highest volumes (27.3 sets per week, on average) gained more strength in 7 out of 9 comparisons, and gained more muscle in 8 out of 9 comparisons, versus the groups training with lower volumes (19 sets per week, on average). Furthermore, all three instances where higher volumes didn’t lead to larger strength gains or more muscle growth occurred in studies on untrained lifters. In the studies on trained lifters, we see more muscle growth and larger strength gains across the board, and the additional strength gains associated with higher volumes (+5.88%, on average) were slightly larger than the additional muscle growth associated with higher volumes (+3.56%, on average). Finally, in 8 out of 9 comparisons, the hypertrophy and strength results were in directional agreement (i.e., the level of volume that led to larger strength gains also led to more muscle growth).

So, what have we seen here?

When we analyze studies on trained lifters that include paired hypertrophy and strength measurements (hypertrophy measurements in muscles that contribute to performance in the tests used to assess strength):

1. It no longer appears that gains in strength plateau at relatively low levels of volume.
2. Relative gains in strength, in general, exceed gains in muscle size at all levels of volume measured.
3. In 31 out of 35 comparisons of paired strength and hypertrophy measurements within a single group of subjects, gains in strength exceeded gains in muscle size. Furthermore, in high-volume studies (studies where the average volume exceeded 20 sets per week), gains in strength exceeded gains in muscle size in 9 out of 10 comparisons.
   1. These patterns also held true when examining whether the groups of subjects training with the highest volumes gained more muscle and/or strength than subjects training with slightly lower volumes within the same study.
4. The gap between gains in strength and gains in muscle size doesn’t appear to get smaller as training volumes increase in these studies, further supporting the idea that increases in volume have a positive impact on both hypertrophy and strength gains.

Different lenses

I want to be entirely clear that I don’t think the analysis presented in the prior section conflicts with the Pelland meta-regression in any substantive way, nor do I think there are any notable problems with the Pelland meta-regression. Rather, the analysis presented above and the Pelland meta-regressions are different analyses addressing slightly different questions. The Pelland meta-regressions are concerned with a) the general relationship between training volume and strength gains, and b) the general relationship between training volume and muscle growth, as observed in the totality of the scientific literature so far. The analysis presented above is concerned with whether the relationship between muscle growth and strength gains is impacted by differing levels of training volume in the specific set of studies most amenable to addressing that specific question.

In other words, studies that do assess strength gains but don’t assess muscle growth do tell you something about the general relationship between volume and strength gains. However, they don’t tell you much about the interplay between hypertrophy and strength gains at differing levels of training volume.

Furthermore, when discussing this topic with two of the authors (Josh Pelland and Zac Robinson), they pointed out a significant reason why strength gains may have appeared to plateau at lower levels of volume in their analysis: in the research we have thusfar, training specificity tends to be lower in the studies testing higher levels of training volume.

It’s well-established that specificity is important for strength gains. This applies to both exercise specificity and loading specificity. If you want to improve your squat strength, a set of squats is generally more productive than a set of knee extensions, for example. Similarly, if you want to improve your 1RM, training with a relatively high load is generally more productive than training with a relatively low load.

So, with that in mind, they calculated the composite specificity of all of the studies testing the impact of training volume on strength gains. Composite specificity was the average of two metrics:

1. Load specificity: (rep max tested ÷ rep range trained) × 100
2. Exercise specificity: (total sets of the exercise tested ÷ total sets training the prime mover for that exercise) × 100

They found that specificity was generally lower in the studies testing higher levels of training volume:

Training specificity for strength assessments was generally lower in the higher-volume studies included in the Pelland meta-regressions.

Just to head off a potential criticism, I don’t think this decrease in specificity at higher levels of volume weakens Pelland’s volume vs. strength meta-regression in the slightest. Rather, I think it illustrates an important thing to keep in mind when digesting any piece of meta-science: “The” effect size does not exist (or, in this case, “the” dose-response relationship does not exist). Studies are non-random, but they tend to be non-random in fairly logical ways. However, meta-analyses or meta-regressions are fundamentally descriptive pieces of research, so their results will necessarily reflect this non-randomness: they (ideally) reflect some true underlying effect or relationship, but they also necessarily reflect the specific study design decisions that researchers have made.

In this case, low-volume and high-volume studies that assess strength gains tend to differ in systematic but ecologically valid ways: in the “real world,” low-volume training also tends to be more specific to strength adaptations than higher-volume training. If you’re training with low volumes, and you care about strength gains at all, you’re probably going to be training with pretty high specificity by using relatively heavy loads, and primarily focusing on the small handful of exercises for which you’d like to improve your performance. Conversely, if you’re training with higher volumes, you’re probably someone who values both strength and hypertrophy (so you probably won’t be training with loads that only have high specificity for strength gains), and you probably aren’t going to be getting all of your weekly volume for a particular muscle group from a single exercise. These differences are largely reflected in the research assessing strength gains at different levels of training volume: lower-volume studies tend to have higher specificity, and higher-volume studies tend to have lower specificity.

In other words, people who do 5 sets of quad training per week may, on average, increase their squat 1RM by roughly the same amount as people who do 15 sets of quad training per week over a 2-3 month period (which is what’s reflected in the Pelland meta-regression). However, that doesn’t necessarily imply that 5 sets and 15 sets of quad training have the same causal impact on strength gains. Rather, (I believe) it implies that the beneficial impact of higher volumes can counterbalance the negative impact of lower specificity: a lot of the people who only do 5 sets of quad training per week have a relatively low training status, and are just doing 5 sets of heavy squats per week, whereas the people who do 15 sets of quad training per week generally have more prior training experience, and they’re doing more exercises and training with slightly lower relative loads. But, that doesn’t mean that 5 weekly sets of heavy squats are just as effective for gaining strength (or building muscle) as 10 or 15 sets of heavy squats would be.

Revisiting our arguments

If you’ll recall from the introduction, these are the arguments that lead to the conclusion that higher training volumes don’t actually result in more hypertrophy:

Argument 1:

Premise 1: Muscular strength capacity is fully determined by the number of parallel actin/myosin crossbridges aligned with the muscle’s line of pull.

Premise 2: Myofibrillar hypertrophy is characterized by an increase in the total number of parallel actin/myosin crossbridges in a muscle.

Conclusion 1: Therefore, myofibrillar hypertrophy increases muscular strength capacity, and an increase in muscle size that’s not accompanied by an increase in strength capacity cannot reflect myofibrillar hypertrophy.

Argument 2:

Premise 1: Myofibrillar hypertrophy increases muscular strength capacity, and an increase in muscle size that’s not accompanied by an increase in strength capacity cannot reflect myofibrillar hypertrophy (the conclusion of Argument 1 is a premise in Argument 2).

Premise 2: Higher training volumes fail to cause larger strength gains when volumes exceed ~5 sets per week.

Conclusion 2: Therefore, higher training volumes (past ~5-10 sets per week) cannot actually be promoting greater myofibrillar hypertrophy.

Inference flowing from Conclusion 2: The apparent increases in hypertrophy at higher training volumes must therefore be the result of muscle swelling, edema, or sarcoplasmic hypertrophy.

This section of the article addressed Argument 2. As we’ve seen, higher training volumes do actually promote larger strength gains. So, even if you believe there are no problems at all with Argument 1 (and I believe there are – that’s the topic of the next major section of this article), we’ve seen that a closer analysis of the strength data would invert the conclusion of Argument 2:

Argument 2 (revised with the understanding that higher training volumes also promote larger strength gains):

Premise 1: Myofibrillar hypertrophy increases muscular strength capacity, and an increase in muscle size that’s not accompanied by an increase in strength capacity cannot reflect myofibrillar hypertrophy.

Premise 2: Higher training volumes lead to larger strength gains.

Tentative Conclusion 2: Therefore, this suggests that higher training volumes also lead to greater myofibrillar hypertrophy (this must be tentative, since observed increases in strength gains are not always reflective of increases in strength capacity for methodological reasons; this will be discussed more later).

Final thoughts on the Pelland meta-regressions

I want to be entirely clear about one point: I think the Pelland meta-regressions are an excellent piece of meta-science. I don’t have any substantive criticisms. Nothing about this article is intended to cast aspersions on the researchers, nor is it intended to imply that they should have analyzed their data differently.

The meta-regressions are descriptive pieces of research. If you read the methods and statistics sections of the paper, you’ll know what studies were included, how the data was analyzed, and why the researchers made the analytical decisions they made (all of which are extremely defensible). But, when a researcher explains to you what data they included, and how they analyzed it, it’s your job as a reader to understand what types of inferences you can and can’t draw from the analysis. When the inferences you’d like to draw from an analysis aren’t the types of inferences the analysis directly supports, you need to do some legwork to determine whether the data actually support the interpretation you’d like to reach, or whether a closer analysis calls your preferred interpretation into question.

So, my gripe is with the ways people have improperly used these meta-regressions to argue for a position that’s not supported by the data (I’d almost go as far as to say it’s an interpretation that’s refuted by the data). I don’t know if it’s due to ignorance, motivated reasoning, or a complete lack of intellectual curiosity, but I expect better from people who claim to understand, synthesize, and communicate research to a lay audience professionally.

Speaking of motivated reasoning…

A simple guide to maximize hypertrophy with zero effort

Let’s briefly forget everything discussed in this section of this article, and return to where we started. 

Let’s again assume that strength data are strongly informative about hypertrophy. Let’s assume that, when further increases in a training variable don’t lead to larger gains in strength, any gains in size are illusory. These apparent increases in size (beyond the point where strength gains plateau) don’t reflect gains in the contractile protein content of the muscle. They merely reflect transient increases in non-contractile elements: sarcoplasmic hypertrophy or muscle edema.

If this is the logical framework you use to make inferences about the optimal way to train for hypertrophy, then I have good news for you: maximizing muscle growth is going to be an absolute breeze.

Since we’ve forgotten everything we previously discussed about training volume and the Pelland meta-regressions, we’re returning to a world where you can maximize strength gains with just 5 sets per week, which means we can maximize muscle growth with just 5 sets per week.

We’re off to a good start. You do still need to lift weights to gain muscle, but you’re not going to need to spend much time at all in the gym. Everyone who’s putting in long hours under the iron is just wasting their time; you know better.

However, maximizing muscle growth is even easier than just dialing back your training volume: you don’t even need to put much effort into your training. In fact, it would be exceedingly hard to put too little effort into your training.

You see, another meta-regression purported to show that training closer to failure results in more muscle growth. But you’re hip to the ways that muscle swelling and edema can pollute so-called “hypertrophy” research. After all, training closer to failure causes more muscle damage (just like pesky training volume), so you know not to take these findings at face value. So, you ask whether the paper also reported the relationship between proximity to failure and strength gains, since you know that strength data will help cut through the noise.

You’re in luck! It did report on strength data. And, just as you suspected, training closer to failure is entirely unnecessary. All of that “hypertrophy” is obviously just edema or sarcoplasmic expansion. After all, the results suggest that training with up to (at least) 10 reps in reserve leads to the same strength gains as training to failure. Since you can maximize your strength gains when training 10 reps from failure, you must be able to maximize your muscle growth with 10 reps from failure as well. After all, how could you possibly be gaining more muscle when training closer to failure if that additional contractile tissue isn’t contributing to additional strength gains?

You couldn’t. It would be illogical to believe such a preposterous idea.

In much the same way that doing 20 sets instead of 5 would be a waste of time for muscle growth (since 5 sets lead to the same strength gains as 20 sets), training to failure instead of keeping 10 reps in reserve would be a massive waste of effort for muscle growth (since training with 10 reps in reserve leads to the same strength gains as training with 0 reps in reserve).

So, there you have it! Following this train of logic to its natural conclusion, you’ve gained an understanding of hypertrophy training most mere mortals could only dream of. You’ve seen through the matrix. You’ve discovered the holy grail: the biggest mistake most people make is trying. If they wanted to maximize their growth, they wouldn’t train so much, and they wouldn’t train so hard. The thing holding them back is effort. Namely, too much effort.

If only they understood that training that maximizes strength (as understood by a surface reading of a few meta-regressions) also maximizes muscle growth, they’d also grasp the sublime truth that you now understand: the optimal training prescription to maximize hypertrophy is 5 sets per week, with 10 reps in reserve.

And now we return to the current timeline.

Obviously the prior fanciful description of our enlightened hypertrophy guru was a bit of transparent shitposting. No one in their right mind believes that 5 sets per week with 10 reps in reserve will maximize hypertrophy. However, it was shitposting with a purpose.

Maybe you didn’t follow my re-analysis of the volume literature. Maybe you think I messed something up. Maybe you still buy into the interpretation of the Pelland meta-regression that would suggest that high volumes are unnecessary to maximize muscle growth because strength gains are maximized with low volumes.

If that’s the case, I’d simply ask you to keep going. Apply your analytical methods consistently. The proximity to failure literature is an exact mirror of the volume literature, as commonly understood through the lens of the most recent meta-regressions on those topics:

1. Strength gains are maximized with low volumes; further increases in volume don’t decrease strength gains, but they don’t increase strength gains either.
   1. Strength gains are maximized at higher RIRs. Training closer to failure doesn’t decrease strength gains, but it doesn’t increase strength gains either.
2. Further increases in muscle growth appear to occur as you increase volume.
   1. Further increases in muscle growth appear to occur as you train closer to failure.
3. However, higher volumes are known to cause more fatigue and more muscle damage.
   1. However, training closer to failure is known to cause more fatigue and more muscle damage.
4. Therefore, training with higher volumes doesn’t actually cause more muscle growth. Any apparent increases in muscle growth are most likely just the result of increases in muscle swelling or sarcoplasmic hypertrophy.
   1. Therefore, training closer to failure doesn’t actually cause more muscle growth. Any apparent increases in muscle growth are most likely just the result of increases in muscle swelling or sarcoplasmic hypertrophy.

So, if this leads you to the understanding that higher volumes aren’t necessary to maximize muscle growth, and the apparent increases in muscle growth are simply the result of swelling, edema, or sarcoplasmic hypertrophy, I simply ask for a bit of consistency.

I’d either like to hear a full-throated endorsement of the idea that the optimal training prescription for maximizing hypertrophy is 5 sets per week with 10 reps in reserve…

Or, I’d ask you to do a bit of soul searching, and realize you don’t actually believe your own argument.

If you believe that training closer to failure does actually lead to more muscle growth, you don’t necessarily need to do a full 180 and endorse the idea that higher training volumes are necessary to maximize muscle growth. But you do need to abandon the argument that the apparent plateau in strength gains at lower training volumes is a strong evidentiary basis for arguing that hypertrophy is maximized with low training volumes. Conversely, if you continue clinging to the argument that additional gains in muscle must contribute to additional gains in strength (and if they don’t, any apparent increases in muscle mass are simply the product of swelling, edema, or sarcoplasmic hypertrophy), the only intellectually consistent position is to abandon your belief that training closer to failure is beneficial for hypertrophy. Otherwise, you’re employing that argument in bad faith, and you (now) know it.

Just something to chew on.

However, you may still be convinced that additional gains in muscle contractile protein content must yield additional gains in strength, such that apparent gains in size without additional strength gains must reveal that the gains in size are actually the result of swelling, edema, or sarcoplasmic hypertrophy. If that’s the case I’d just note that this analytical framework would support the idea that volume is more important for muscle growth than proximity to failure, not the other way around. As discussed previously in this article, the data do suggest that higher training volumes increase both muscle mass and strength, but the same is not true for training close to failure (i.e., even in within-study comparisons, training closer to failure doesn’t typically lead to larger strength gains).

More reason to doubt transitive inferences relating strength data to hypertrophy outcomes

In the first article in this series, I asserted that strength data don’t tell you much about muscle growth, but I didn’t really discuss why. So far in this article, I’ve discussed one of the technical reasons why you generally can’t (or at least shouldn’t) make inferences about hypertrophy from strength data when comparing the results of various meta-analyses or meta-regressions:

Different sets of studies (or at least, sets of studies that only partially overlap) are typically used for analyses of the impact of a particular variable on strength outcomes, and analyses of the impact of a particular variable on muscle growth. So, apparent divergences between strength and hypertrophy outcomes may just be an artifact of the characteristics of the non-overlapping studies (training status, age, exercise selection, methods of measurement, other aspects of program design, etc.).

And, for the record, that is the main reason why I personally think it’s silly to make strong inferences about hypertrophy from strength data most of the time. Individual studies are almost never strongly informative, so a firm understanding of the impact of a particular variable on strength or hypertrophy outcomes typically requires a meta-analysis or meta-regression. However, meta-analyses looking at the impact of a particular variable on hypertrophy, and meta-analyses looking at the impact of the same variable on strength gains, almost inevitably use different sets of studies. There are some hypertrophy studies that don’t measure strength, and there are a lot of strength studies that don’t measure hypertrophy. So, there are always plenty of studies that are included in hypertrophy meta-analyses and excluded from strength meta-analyses, and vice versa. As a result, you’re almost always making an apples-to-oranges comparison when you try to see how well the results of a hypertrophy meta-analysis line up with the results of a strength meta-analysis. Unless you have the time (and interest, and skillset, and journal access) to re-extract the data and re-run a bunch of analyses, any apparent similarities or differences may be the result of a direct impact (or lack of impact) of hypertrophy on strength outcomes, but it’s far more likely to just be the result of sampling variance, slightly different populations or experimental protocols in the studies, etc.

However, this doesn’t necessarily refute the argument that strength data are informative about hypertrophy outcomes in general. Rather, it could merely suggest that you need to be more careful when selecting studies and analytical techniques for making such an argument. In other words, the argument could be physiologically and mechanistically valid, and the only issue is that the way people typically try to make the argument is statistically invalid.

In fact, it would be very convenient for me to stop here. I do think that training with higher volumes leads to more muscle growth, and I do think that a careful analysis of the data suggests that gains in muscle mass and gains in strength run roughly in parallel at moderate-to-high training volumes. So, it would be quite cozy for me to stop here and claim that my beliefs about the impact of volume on hypertrophy are supported by the research investigating the impact of volume on strength gains.

Unfortunately (for me), I couldn’t make that argument in good faith, even though it would support my biases. Rather, we can’t use strength data to make transitive inferences about hypertrophy for a much more fundamental reason: the argument itself relies on a premise that’s untrue.

At a very basic level, we’re dealing with a deductive argument. In a deductive argument, you should reach correct conclusions if the argument is sound. If the argument is not sound, it should not be expected to reliably reach correct conclusions. Soundness relies on two conditions:

1. The argument is valid in form.
2. The premises of the argument are true.

I won’t discuss formal validity here, because the argument is formally valid (it’s essentially just a basic syllogism). But, I will be drilling down on the requirements a premise needs to meet in order to be able to draw causal inferences about hypertrophy from strength data.

Let’s return to Argument 1 from before:

Premise 1: Muscular strength capacity is fully determined by the number of parallel actin/myosin crossbridges aligned with the muscle’s line of pull.

Premise 2: Myofibrillar hypertrophy is characterized by an increase in the total number of parallel actin/myosin crossbridges in a muscle.

Conclusion 1: Therefore, myofibrillar hypertrophy increases muscular strength capacity, and an increase in muscle size that’s not accompanied by an increase in strength capacity cannot reflect myofibrillar hypertrophy.

As mentioned above, this is a formally valid argument, meaning that the conclusions are true if the premises are true. Furthermore, Premise 2 is true (it straightforwardly follows from the definition of myofibrillar hypertrophy). Premise 1 is the issue.

To explain why, I need to first explain two very important choices in wording: “strength capacity,” and “fully determined.”

I’ll start by explaining the choice of “fully determined” in Premise 1. This wording is critical for the conclusion to be valid. In essence, this wording means Premise 1 is reversible: if you know a muscle’s strength capacity, you necessarily know the number of parallel actin/myosin crossbridges, and if you know the number of parallel actin/myosin crossbridges, you necessarily know the muscle’s strength capacity. Without this wording, you could conclude that myofibrillar hypertrophy should increase a muscle’s capacity for strength, but you couldn’t conclude that a change in the capacity for strength necessarily tells you whether or not myofibrillar hypertrophy has occurred.

Just to illustrate, this is a formally valid argument

Premise 1: All jeans are blue garments.

Premise 2: Joey is wearing jeans.

Conclusion: Joey is wearing a blue garment.

Alternate premise 2: Joey is not wearing a blue garment.

Alternate conclusion: Joey is not wearing jeans.

However, this is a formally invalid argument:

Premise 1: All jeans are blue garments.

Premise 2: Joey is not wearing jeans.

Conclusion: Joey is not wearing a blue garment.

Alternate premise 2: Joey is wearing a blue garment.

Alternate conclusion: Joey is wearing jeans.

In this argument, all jeans are blue, but all blue garments are not necessarily jeans. So, Joey could be wearing a blue garment other than jeans.

In these examples, premise 1 wasn’t reversible, which limits the number of valid inferences we can reach. You know that all jeans are blue (i.e., the type of garment implies the color of the garment), but you don’t know that all blue garments are jeans (i.e., the color of the garment does not necessarily imply the type of garment, and ruling out one type of garment does not necessarily rule out the color of the garment). However, if we make premise 1 reversible, the previously invalid argument above now becomes valid:

Premise 1: All jeans are blue garments and all blue garments are jeans.

Premise 2: Joey is not wearing jeans.

Conclusion: Joey is not wearing a blue garment.

So, returning to the argument about the relationship between strength and hypertrophy, inferences about parallel crossbridges and muscular strength capacity must be reversible. If this premise isn’t reversible – the number of parallel crossbridges directly contributes to strength capacity, but strength capacity isn’t merely reflective of the number of parallel crossbridges (i.e., because other factors contribute to strength capacity) – you couldn’t conclude that an increase in strength capacity necessarily reflects an increase in parallel crossbridges, or that the lack of increase in strength capacity necessarily reflects a lack of increase in parallel crossbridges.

Moving on, I need to briefly explain “strength capacity” of the muscle. “Capacity,” in this case, refers to the maximum amount of contractile force that could be generated by the muscle itself. For example, a 15-gallon fuel tank has a greater capacity than a 10-gallon fuel tank. That doesn’t necessarily mean that the 15-gallon tank currently has more fuel in it, but it could hold more fuel. I refer to “capacity” here because a simple assessment of strength at a particular moment in time may not reflect the muscle’s true capacity to generate force due to poor technique, lack of motivation, poor coordination, etc. Notably, divergences between assessments of muscle strength and muscle strength capacity are primarily due to factors originating outside of the muscle. The muscle could generate a certain amount of force, but it’s generating less than its potential maximum due to insufficient activation or coordination of its motor units and muscle fibers.¹

I realize that was a long introduction, so let’s return to the argument at hand to recenter ourselves:

Premise 1: Muscular strength capacity is fully determined by the number of parallel actin/myosin crossbridges aligned with the muscle’s line of pull.

Premise 2: Myofibrillar hypertrophy is characterized by an increase in the total number of parallel actin/myosin crossbridges in a muscle.

Conclusion 1: Therefore, myofibrillar hypertrophy increases muscular strength capacity, and an increase in muscle size that’s not accompanied by an increase in strength capacity cannot reflect myofibrillar hypertrophy.

As mentioned previously, I believe that premise 1 is false, which means the argument is not sound, and therefore, the conclusion is not necessarily true.

If premise 1 were true, then changes in whole-muscle strength capacity should directly scale with changes in single-fiber strength capacity. In other words, if muscle fibers get 10% larger and 2% stronger per unit of cross-sectional area, the individual fibers are now 12.2% stronger, so the muscle’s strength capacity should also increase by 12.2%. And, notably, if strength capacity is fully determined by the number of parallel actin/myosin crossbridges, if fibers increase in strength by 12.2%, the whole muscle cannot increase in strength capacity by more than 12.2%. If strength capacity of the whole muscle increased by more than 12.2%, that would mean that something other than the number of parallel actin/myosin crossbridges influences strength capacity, invalidating premise 1, and thus, invalidating the conclusion.

Just as a terminology note, maximal muscle fiber force per unit of cross-sectional area is typically referred to as specific force or specific tension, and maximal whole-muscle force per unit of physiological cross-sectional area is typically referred to as specific force or normalized muscle force.

In essence, my argument is that if premise 1 in the prior argument is true, then fiber specific force should determine whole-muscle specific force. Whole-muscle specific force should not exceed fiber specific force (as that would imply that something other than the contractile forces generated by the fibers are contributing to the total whole-muscle force generated at the tendon), and changes in whole-muscle specific force should not differ from changes in fiber specific force (changes in fiber specific force are determined by the number of parallel actin/myosin crossbridges, so if changes in whole-muscle specific force differ from changes in fiber specific force, that implies that changes in the number of parallel actin/myosin crossbridges aren’t the only thing that influences a muscle’s force capacity).

To start with, it doesn’t even seem like you can extrapolate whole-muscle force capacity from single-fiber strength at baseline, even before considering how both change with training.

Human muscle fibers tend to produce between 10-25 Newtons of force per square centimeter of cross-sectional area (averaging around 15.5N/cm² for type I fibers, and around 18N/cm² for type II fibers). Whole muscles, on the other hand, have been reported to produce anywhere between 2-73N/cm², with a median of 26.8N/cm², and an interquartile range of 20-43N/cm². So, the range of normalized whole-muscle forces is considerably wider than the range of normalized fiber forces, and typical normalized whole-muscle forces are greater than typical normalized fiber forces. Both of these facts at least suggest that factors outside of muscle fiber force (i.e., the number of parallel actin/myosin crossbridges) influence muscular strength capacity.

However, to play the skeptic, it’s entirely possible that both of those findings are simply due to measurement error. Measuring the force capacity of single fibers is technically demanding but fairly straightforward (you need to isolate a single muscle fiber from a biopsy, attach it to a super sensitive load cell, and then bathe it in a solution that forces it to contract with maximal force); however, the process of preparing fibers for force measurements causes some fiber swelling. Researchers correct for this swelling, but it’s possible that they’re under-correcting, such that single-fiber specific force in vivo is a bit higher than the literature reports. 

Measuring the force capacity of entire muscles is considerably more challenging: you need to account for fiber pennation angles, coactivation of antagonists, muscle internal moment arms, conversion of tendon forces to muscle forces, etc. I don’t want this to turn into an article on research methods, so I’ll spare you all of the gory details, but there is plenty of room for measurement error when assessing whole-muscle specific force (however, it is important to note that these assessments do involve electrical stimulation of the motor nerve to ensure the muscle achieves full activation and maximal force output, so that the measurement isn’t impacted by variations in motivation, coordination, “neural adaptations,” or capacity to achieve high levels of voluntary activation).

So, baseline measurements in single-fiber versus whole-muscle strength capacity at least suggest that whole-muscle strength capacity isn’t fully determined by single-fiber strength capacity (and, by extension, the number of parallel actin/myosin crossbridges). However, to be sure, we need to see how fiber force capacity and whole-muscle force capacity change with training.

A 2019 meta-analysis purported to examine changes in single-fiber and whole-muscle strength with resistance training. It found that measures of hypertrophy were similar at the whole-muscle and single-fiber level (+4.6% vs. +7%; p = 0.88), but strength gains were much larger at the whole muscle-level (+43.3% vs. +19.9%; p<0.001). And I wish I could stop there, because this seems like a slam dunk. We have a meta-analysis showing that muscle force increases to a much larger extent than single-fiber force following resistance training, which means that muscular strength capacity isn’t fully determined by the number of parallel actin-myosin crossbridges … right?

Unfortunately, the inclusion criteria of that meta-analysis were a bit too broad. Again, I’ll spare you the methodological details, but a simple strength assessment is insufficient for assessing changes in normalized muscle force. Most of the studies in that meta-analysis used strength assessments where gains in technical proficiency and “neural adaptations” could have influenced results. As mentioned previously, to measure strength capacity, you need to use a very simple exercise (typically a single-joint isometric exercise), and electrically stimulate the motor nerve to ensure full activation of the muscle.

There is, to my knowledge, only one applicable study on the topic. I checked the reference lists from both of the most recent meta-analyses on related topics (one, two) and scoured PubMed for a couple of hours, and couldn’t turn up any other papers, unfortunately. However, even though we do only have one applicable study, the research team conducting the studies is one of the teams that performs assessments of normalized muscle force with the highest methodological quality.

In this study by Erskine and colleagues, subjects completed nine weeks of knee extension training. Before and after the training intervention, researchers assessed changes in quadriceps physiological cross-sectional area, specific force, and specific power, along with changes in quadriceps muscle fiber cross-sectional area, specific force, and specific power (specific power is similar to specific force: power output normalized to muscle size).

Gains in physiological cross-sectional area were similar to gains in fiber cross-sectional area (+7.2% vs. +7.8%). However, gains in maximal quadriceps contractile force exceeded gains in maximal contractile force of isolated fibers (+24.1% vs. +16.4%). Consequently, gains in specific force of the quadriceps comfortably exceeded gains in specific force of isolated fibers (+16.2% vs. +6.2%).

So, it certainly appears that something influences muscular strength capacity beyond the number of parallel actin/myosin crossbridges (otherwise, gains in specific force of the muscle should not exceed gains in specific force of isolated fibers). And keep in mind, this result can’t be chalked up to “neural adaptations,” since the method of assessing whole-muscle force involved stimulating the motor nerve to ensure that maximal contractile force was achieved.

The assessment of specific power offers a clue: the researchers found that quadriceps specific power didn’t increase with training. In fact, it nominally (non-significantly) decreased.

At first, this may seem confusing. If specific force increases, you should generally expect specific power to increase as well. After all, power = force × velocity, so if a muscle can produce more force per unit of size, you’d expect that it should also be able to produce more power per unit of size…unless the gains in specific force are offset by losses in shortening velocity.

The researchers propose that the most likely explanation for these findings is that training caused an increase in the number of connections between the muscle fibers and their surrounding connective tissue. When muscle fibers contract, they transmit most of their force laterally into the connective tissue adjacent to the fiber. This connective tissue is continuous with the muscle’s tendons, and it’s the summated forces generated at the tendon that ultimately determine the contractile force that reaches the bone.

An increase in the number of connective tissue attachments would accomplish two purposes. First, it would increase the efficiency of force transmission, thus increasing muscle specific force independent of changes in fiber specific force. Second, it would reduce the effective length of each muscle fiber, thus reducing maximum shortening velocity. The net effect: gains in muscle specific force without gains in muscle specific power, matching the observations of this study.

If this is a difficult concept to grasp, the gears in your car provide a pretty good analogy. Having more connective tissue attachment points is like being in a low gear.

When your engine revs up in a low gear, the car doesn’t go fast, but it can generate a lot of torque, allowing you to climb steep hills or tow a heavy load. Similarly, if you have more connective tissue attachment points, when your brain sends a signal for a maximal muscle contraction, your muscles achieve a slightly lower shortening velocity, but they can generate a lot of contractile force, allowing you to lift heavier loads.

Conversely, when your engine revs up in a high gear, it doesn’t generate as much raw torque, but it does let you cruise down the road at high speeds. Similarly, if you have fewer connective tissue attachment points, when your brain sends a signal for a maximal muscle contraction, your muscles don’t generate quite as much raw contractile force, but they are able to achieve a higher maximum shortening velocity.

This is a new idea to a lot of lifters and coaches, but it’s an old idea in the scientific literature. In fact, it was clearly articulated in a review paper all the way back in 1989. The importance of lateral force transmission is pretty uncontroversial with muscle physiologists, but it’s notoriously difficult to measure in vivo, so there’s not a ton of direct research on how it changes in response to resistance training.

However, we do know that resistance training increases the concentrations of proteins that are present at these attachment points, and we also observe that genotype differences related to these proteins can influence whole-muscle specific force. So, we do have some human data supporting this mechanism for specific force increases with resistance training (in addition to the fairly strong indirect support provided by functional changes in specific force and specific power with resistance training).

So, I personally think there’s sufficient reason to believe that these connective tissue adaptations provide us with a clear mechanism by which a muscle’s strength capacity can change independent of changes in the number of parallel actin/myosin crossbridges aligned with a muscle’s line of pull. As such, let’s return to the argument at hand one last time:

Premise 1: Muscular strength capacity is fully determined by the number of parallel actin/myosin crossbridges aligned with the muscle’s line of pull.

Premise 2: Myofibrillar hypertrophy is characterized by an increase in the total number of parallel actin/myosin crossbridges in a muscle.

Conclusion 1: Therefore, myofibrillar hypertrophy increases muscular strength capacity, and an increase in muscle size that’s not accompanied by an increase in strength capacity cannot reflect myofibrillar hypertrophy.

For the reasons discussed in this section, I believe premise 1 is false: muscular strength capacity is not fully determined by the number of parallel actin/myosin crossbridges aligned with a muscle’s line of pull. Instead, muscular strength capacity is influenced by other factors (namely, connective tissue adaptations that influence the efficiency of lateral force transmission). As such, the second half of the conclusion is not necessarily true.

Therefore, if a training intervention does not lead to larger strength gains, you cannot necessarily conclude that it also fails to cause more hypertrophy (and conversely, if a training intervention does lead to larger strength gains, you cannot necessarily conclude that it does cause more hypertrophy). 

Notably, what we’ve discussed in this section provides us with a potential explanation for why a particular training variable may have divergent effects on hypertrophy and strength adaptations.

If we return to the proximity to failure literature, we observe that training closer to failure does lead to more muscle growth, but it doesn’t lead to larger strength gains. This could suggest that training closer to failure is just causing more muscle edema or sarcoplasmic hypertrophy. However, it could also suggest that training closer to failure just causes smaller connective tissue adaptations to enhance lateral force transmission.

Just to illustrate, if training at 10 reps from failure causes muscle size to increase by 5% due to myofibrillar hypertrophy (i.e., the number of parallel actin/myosin crossbridges remains constant as muscle size increases), and it also causes connective tissue adaptations that would increase specific force by 10%, then training at 10 reps from failure would increase strength by about 15%. Conversely, if training to failure causes muscle size to increase by 10% due to myofibrillar hypertrophy, and it also causes connective tissue adaptations that would increase specific force by 5%, then training to failure would also increase strength by about 15%. In both cases, increases in parallel actin/myosin crossbridges are directly contributing to strength gains, but they don’t fully determine strength gains, because another variable is also influencing the muscle’s strength capacity.

tl;dr: Connective tissue adaptations can increase lateral force transmission, and thus impact gains in muscular strength capacity independent of changes in parallel actin/myosin crossbridges. So, there’s not only a statistical reason why it’s typically dumb to draw inferences about hypertrophy from strength data; there’s also a mechanistic, physiological explanation for why strength and hypertrophy outcomes can diverge.

It’s time to enter the speculation zone

There’s at least some reason to expect that effective hypertrophy training may, at least temporarily, reduce the number of attachments between muscle fibers and their surrounding connective tissue (and thus lead to smaller strength gains than you’d anticipate in the short-to-medium term).

In general, effective hypertrophy training is hard training. Higher training volumes cause more stress than lower training volumes. Training closer to failure causes more stress than training further from failure. Training at longer muscle lengths generally causes more stress than training at shorter muscle lengths. 

While chronic resistance training probably increases the number of attachment points between muscles and their surrounding connective tissues, an acute, stressful bout of resistance training likely decreases the number of connections. Following an unaccustomed tension stimulus, human subjects experience upregulation of a protein called tenascin C (or TNC for short). TNC interferes with adhesions between muscle fibers and the surrounding extracellular matrix, which is thought to contribute to the large decrease in strength new lifters experience for several days following a workout. However, this deadhesion process is thought to be necessary for tissue repair and regeneration. It also protects muscle fibers and the extracellular matrix from additional damage, since muscle contractions will cause less shear stress between the muscle fibers and the surrounding endomysium. Following this deadhesion process, connective tissue synthesis rates increase, helping to repair and strengthen the extracellular matrix.

Of note, all of these processes are attenuated with repeated training bouts: smaller increases in TNC, smaller additional increases in connective tissue synthesis, etc. And, in general, that’s a good thing. Your connective tissue is strong enough to withstand the forces it’s exposed to during training, and your muscle fibers experience less damage from training, so you don’t need to have as much deadhesion to allow for muscle fiber and connective tissue repair, and you don’t need connective tissue synthesis rates to increase quite as much.

However, connective tissue adaptations tend to run in parallel with muscle fiber adaptations – as fibers get bigger and stronger, the connective tissue matrix also needs to get stronger in order to withstand the additional forces that the fibers can create. So, I think it’s plausible that training that’s sufficiently challenging to cause continued muscle growth is also training that causes enough stress to promote continued connective tissue adaptations. Or, potentially, training that’s not sufficiently challenging to cause continued connective tissue adaptations isn’t sufficiently challenging to cause continued muscle growth, since connective tissue adaptations may actually place constraints on muscle growth – if your connective tissue isn’t strong enough, that may pump the brakes on muscle growth. One review paper even suggested that extracellular matrix turnover may be “a rate-limiting step during load-induced hypertrophy.” For a fun rabbit hole that goes far beyond the scope of this article, you may enjoy reading about TGF-β receptor regulation, and its impact on connective tissue deposition and hypertrophy (even in the absence of a significant tension stimulus).

So, if your muscles aren’t growing, it’s possible that they also don’t need to undergo further connective tissue adaptations, which means they don’t need to undergo much (if any) deadhesion to allow for muscle and connective tissue repair and adaptations. As a result, you have more connections between your muscle fibers and the surrounding connective tissue matrix at any given moment in time, thus increasing your muscles’ strength capacity.

However, if your muscles are growing, that could mean that they’re simultaneously experiencing parallel connective tissue adaptations, which means they do need to undergo more deadhesion to allow for muscle and connective tissue repair and adaptations. As a result, you have fewer connections between your muscle fibers and the surrounding connective tissue matrix at any given moment in time, thus (temporarily) decreasing your muscles’ strength capacity. Furthermore, if deadhesion scales with the necessary rate of connective tissue adaptation, and the necessary rate of connective tissue adaptation scales with the rate of muscle growth, this could also mean that faster rates of hypertrophy are associated with greater deadhesion, and larger (temporary) reductions in your muscles’ strength capacity.

To be entirely clear, this is all speculative. Furthermore, if this effect exists, I don’t think it’s huge. For the most part, I think it would account for shifts in specific force of about ±5% (which is comfortably within the range of normal whole-muscle specific force changes with training). However, this effect wouldn’t need to be particularly large to account for apparent divergences between strength and hypertrophy results in the literature.

To illustrate, the Robinson proximity to failure meta-analysis found that, with anywhere from 0-10 reps in reserve, strength tended to increase by around 15%, on average. Furthermore, it found that hypertrophy results decreased at greater proximities to failure, from ~10% increases in muscle size at 0RIR, to ~5% increases in muscle size at 10RIR. So, if this “divergence” was fully explained by differential changes in whole-muscle specific force, that would imply that training with 10RIR led to a 10% increase in specific force, and training with 0RIR led to a 5% increase in specific force, as illustrated in the figure below.

For what it’s worth, that feels extremely reasonable to me.² I can easily buy into the idea that more challenging training builds more muscle, but it also means your connective tissue is experiencing a greater rate of turnover, such that a small handful of potential connections between the muscle fibers and endomysium aren’t “hooked up” at any given point in time.

And, also for what it’s worth, this would also provide a tidy explanation for the increase in strength performance you experience when shifting from a more challenging block of training to a less challenging block of training (from higher to lower volume, or just when tapering for a 1RM test). There’s surprisingly little evidence for the idea that fatigue truly “accumulates” during resistance training, but many people have the experience of fatigue seeming to accumulate during a challenging block of training, and dissipating when they pull back a bit. However, I think it’s at least possible that connective tissue strain does accumulate, leading to some deadhesion between muscle fibers and the surrounding connective tissue, decreasing specific force. That’s how I personally feel near the end of a challenging training block. It doesn’t really feel like I’m struggling to exert a maximal effort or recruit motor units – it just feels like my muscles are still working hard, but aren’t producing quite as much force as I expect. Then, when training loads decrease, the connective tissue can recuperate, leading to an increase in adhesions between the muscle fibers and connective tissue matrix, and consequently, a recovery of force output.

Furthermore, there’s some indirect longitudinal evidence that (I think) supports this hypothesis.

In a 2018 study by Bjørnsen and colleagues (which we previously wrote about here), untrained subjects completed two intensive blocks of training. Each block of training lasted for five days. During the first three days, subjects trained their quads once per day. During the last two days, subjects trained their quads twice per day. Each session consisted of 4 sets of knee extensions to failure with blood flow restriction, utilizing a load equal to 20% of 1RM. So, each five-day block consisted of 28 sets of quad training to failure, with no rest days. The two blocks were separated by 10 days of rest.

Quadriceps (vastus lateralis and rectus femoris) cross-sectional area was assessed on every training day, during the middle of the rest week, and at 3 and 10 days after the second block of training. Knee extension 1RM was assessed pre-training, during the middle of the rest week, and at 3, 10, and 20 days after the second block of training.

As you’d expect, quadriceps CSA rapidly increased during the first block of training, likely primarily due to increases in muscle swelling. This was accompanied by large increases in blood markers of muscle damage (creatine kinase and myoglobin). Then, during the 10 days of rest, quadriceps CSA decreased, settling at a ~2-3% increase over baseline before the next block of training started.

During the second block of training, quad CSA again increased rapidly. But this time, it didn’t dive back toward baseline nearly as quickly. On the last day of the training program, quad CSA was about 7-8% greater than baseline. After 10 days of rest, quad CSA settled at a 6-7% increase from baseline. This second block of training wasn’t accompanied by increased markers of muscle damage, and 10 days is plenty of time for any potential edema to subside. So, we can be pretty confident that these subjects truly experienced a 6-7% increase in quad size.

Strength measures, on the other hand, tell a different story. During the rest week, subjects were about 2-3% weaker than they were pre-training. Three days after the final training session of the second block, subjects’ knee extension 1RM strength was still about 2-3% below baseline. But after 10 days of rest, their strength had recovered to about 2% above baseline. After 20 days of rest, their 1RM knee extension strength was about 6-7% above baseline.

If the strength results were merely due to “fatigue dissipating,” 10 days of rest is plenty of time for fatigue to dissipate. Instead, I think it’s far more likely that it took about a month for their muscles to finish off the connective tissue adaptations that started during the first block of training (extracellular matrix restructuring is still taking place up to at least 4 weeks after a single workout in untrained subjects). While those adaptations were proceeding, their specific force was depressed. Once those adaptations were complete and molecular signals encouraging deadhesion went away, specific force could recover, yielding strength adaptations that ran in parallel with gains in quadriceps size.

I also think this study provides a good object lesson in the downsides of assuming that apparent divergences in strength and hypertrophy outcomes are actually divergences, instead of noisy reflections of your most recent block of training. If the final assessment of strength and hypertrophy took place three days after the last training session (which is pretty typical), we would have seen that quad strength only increased by about 2%, while quad CSA increased by about 6-7%. I’m sure that many people would look at that result, assume that gains in size can’t outpace gains in strength, and conclude that most of the “hypertrophy” observed was just due to muscle swelling, edema, or sarcoplasmic hypertrophy. However, with a bit more time, we can see that muscle CSA didn’t drop back toward baseline much (if any), suggesting that the observed 6-7% increase in CSA was truly reflective of actual hypertrophy. Furthermore, we can see that strength continued increasing over the next couple of weeks, eventually landing on a value that closely mirrored the observed increases in quad CSA (thus suggesting that the observed hypertrophy was reflective of actual myofibrillar hypertrophy).

Before moving on, I just want to reiterate that this section is extremely speculative. As mentioned previously, there’s not a ton of data examining changes in muscle fiber/endomysium connections with longitudinal resistance training. So, I think these adaptations would provide a tidy explanation for several phenomena, and I personally believe that intramuscular connective tissue adaptations are far more important than most people realize. However, I will readily acknowledge that this section of the article was at least as much theorycrafting as it was research analysis. If it rings true to you, great. If not, that’s also fine.

Furthermore, I also want to note that I don’t think divergent connective tissue adaptations are the only reason that strength and hypertrophy results might diverge. I simply wanted to drill down into one example of an adaptation that influences strength capacity independent of hypertrophy.

A brief recap

The next section of this article will discuss whether higher training volumes are actually just contributing to muscle swelling, rather than causing more actual hypertrophy. But, before diving into that section, I think it’s worth briefly recapping what we’ve covered so far.

It’s my contention that higher training volumes do actually promote greater muscle growth. The evidence supporting this position is direct and straightforward: we generally observe more hypertrophy when research participants train with higher volumes. We observe this relationship when comparing between studies (subjects in studies that use higher volumes tend to grow more than subjects in studies that use lower volumes), and when comparing within studies (in studies that test multiple levels of training volume, groups training with higher volumes tend to grow more than groups training with lower volumes).

However, there’s a popular position contending that higher volumes (past a fairly low point – somewhere around 5-10 sets per week) don’t actually lead to more muscle growth. This position is supported by the belief that strength gains plateau at a fairly low level of volume. It is then argued that if higher volumes led to more muscle growth (myofibrillar hypertrophy), they’d also lead to larger strength gains – since higher volumes don’t lead to larger strength gains, they must not actually be causing more muscle growth. Therefore, the apparent increases in muscle growth at higher training volumes must be the result of increases in muscle size that aren’t the result of myofibrillar protein accretion (i.e., they simply reflect muscle swelling or sarcoplasmic hypertrophy).

To this point in the article, we’ve seen that the arguments against the hypertrophic effects of higher volumes fall apart under scrutiny.

1. Higher training volumes do actually promote larger strength gains.
   1. When examining paired strength and hypertrophy assessments in trained lifters (i.e., if a study assesses changes in quad size, it also assesses changes in quad strength), we observe that higher training volumes promote larger gains in size and strength.
   2. The apparent “plateau” in strength gains at 5 sets per week primarily comes from studies in untrained lifters.
   3. When assessing within-study differences in strength gains at escalating levels of training volume, we reliably observe larger strength gains at higher levels of training volume.
2. Even if higher training volumes didn’t lead to larger strength gains, that would not necessarily imply that higher training volumes don’t lead to more muscle growth.
   1. Muscular strength capacity is not fully determined by the number of parallel actin/myosin crossbridges. Other structural adaptations (not just improvements in coordination, motor skill, or “neural adaptations”) can also influence strength, meaning that whole-muscle strength adaptations are not merely reflective of fiber-level adaptations.

So, I could stop there. The primary evidentiary basis for the argument against higher training volumes (the apparent strength plateau at low levels of training volume) is mostly a mirage. And, even if it wasn’t, the primary argument against higher training volumes is not a sound argument (it’s formally valid, but it relies on false premises):

I’ll also note, refuting either Premise 1 of Argument 1 or Premise 2 of Argument 2 is sufficient to refute Conclusion 2.

If you’re unconvinced by the section of this article about connective tissue adaptations (Premise 1 of Argument 1), but you were convinced by the section about the impacts of training volume on strength gains (Premise 2 of Argument 2), then this line of argumentation still holds up for you, but it now results in further evidence in favor of higher training volumes increasing muscle growth:

Partial Disagreement A: You were unconvinced by my refutation of Premise 1 of Argument 1, but you were convinced that higher training volumes lead to larger strength gains

Argument 1:

Premise 1: Muscular strength capacity is fully determined by the number of parallel actin/myosin crossbridges aligned with the muscle’s line of pull.

Premise 2: Myofibrillar hypertrophy is characterized by an increase in the total number of parallel actin/myosin crossbridges in a muscle.

Conclusion 1: Therefore, myofibrillar hypertrophy increases muscular strength capacity, and an increase in muscle size that’s not accompanied by an increase in strength capacity cannot reflect myofibrillar hypertrophy.

Argument 2:

Premise 1: Myofibrillar hypertrophy increases a muscle’s strength capacity.

Premise 2: Higher training volumes lead to larger strength gains.

Tentative Conclusion 2: Therefore, this at least suggests that higher training volumes also lead to greater myofibrillar hypertrophy (this must be tentative, since observed increases in strength gains are not always reflective of increases in strength capacity for the methodological reasons discussed previously).

Similarly, if you were convinced by the section of this article about connective tissue adaptations (Premise 1 of Argument 1), but you weren’t convinced by the section about the impacts of training volume on strength gains (Premise 2 of Argument 2), the conclusion of Argument 2 still isn’t a sound conclusion, since the unsound conclusion of Argument 1 is carried over as a faulty premise into Argument 2:

Partial Disagreement B: You were unconvinced by my refutation of Premise 2 of Argument 2, but you were convinced that muscular strength capacity is not fully determined myofibrillar hypertrophy

Argument 1:

Premise 1: Muscular strength capacity is fully determined by the number of parallel actin/myosin crossbridges aligned with the muscle’s line of pull.

Premise 2: Myofibrillar hypertrophy is characterized by an increase in the total number of parallel actin/myosin crossbridges in a muscle.

Conclusion 1: Therefore, myofibrillar hypertrophy increases muscular strength capacity, and an increase in muscle size that’s not accompanied by an increase in strength capacity cannot reflect myofibrillar hypertrophy.

Argument 2:

Premise 1: An increase in muscle size that’s not accompanied by an increase in strength capacity cannot reflect myofibrillar hypertrophy (the faulty conclusion of Argument 1 is a premise in Argument 2).

Premise 2: Higher training volumes fail to cause larger strength gains when volumes exceed ~5 sets per week. (You still take this to be a true premise.)

Conclusion 2: Therefore, higher training volumes (past ~5-10 sets per week) cannot actually be promoting greater myofibrillar hypertrophy (this is still not a sound conclusion, since Premise 1 is faulty). 

So, if you’ve found either main section of this article compelling, you’re left with one of two conclusions:

1. You’ve been convinced that hypertrophy outcomes can’t be inferred from strength data. Therefore, if we want to know how training volume impacts hypertrophy, we should look at the actual hypertrophy data, and the hypertrophy data suggests that higher training volumes increase hypertrophy.
2. You still believe that hypertrophy outcomes can be inferred from strength data. But, you’ve been convinced that the strength data suggest that higher training volumes increase strength gains. Therefore, the strength data corroborate the hypertrophy data suggesting that higher training volumes increase hypertrophy.

So, for most readers, I shouldn’t need to address the contention that higher training volumes actually just increase muscle swelling, rather than increasing hypertrophy. The arguments and affirmative evidence supporting the position that higher training volumes don’t lead to more muscle growth are lying in tatters on the ground. So, the only reasonable position is that higher training volumes appear to lead to more muscle growth because they actually do lead to more muscle growth: that’s what the direct longitudinal human evidence suggests, and there are no strong arguments or evidence to the contrary.

But, I’m sure there are still some skeptics. And for everyone else who’s already on board, addressing contentions about muscle swelling should provide you with even more confidence that higher training volumes are actually causing more real hypertrophy. So, let’s dive in.

Is it all just a matter of swelling?

To start this section off, I’ll just lead with the most damning admission for my position: We have no affirmative evidence demonstrating that higher training volumes don’t cause more post-workout muscle swelling that persists over the entire course of an 8-12 week training program.

However, I’ll follow that up by noting the inverse is also true: there’s not affirmative evidence demonstrating that higher training volumes do cause more post-workout muscle swelling that persists over the entire course of an 8-12 week training program.

So, we’re necessarily dealing with a probabilistic argument: is it more likely that post-workout muscle swelling explains all (or at least most) of the apparent impact of higher training volumes on hypertrophy? Or is it more likely that post-workout muscle swelling is attenuated over time, such that post-workout muscle swelling is unlikely to meaningfully influence the apparent impact of higher training volumes on hypertrophy?

The case for swelling

The argument that muscle swelling is the cause for the apparent increase in hypertrophy with higher training volumes is fairly straightforward:

After a workout, our muscles experience some temporary swelling. The two main factors contributing to this swelling are:

1. Disruptions in the membrane of the muscle fiber and the activation of various stretch-activated channels allow the inflow of various ions (primarily sodium). This increases the osmotic pressure inside the muscle fiber, causing water to flow in. This is likely the predominant mechanism of muscle swelling for the first 24-48 hours after a workout.
2. The local inflammation caused by muscle damage and oxidative stress leads to the activation and infiltration of immune cells, which release signaling molecules that increase vascular permeability. This helps the muscles clear out waste products and damaged proteins, and receive more fuel and amino acids from circulation, but it also increases the amount of fluid that flows into the muscle (including increased flow into the muscle fibers, and also the extracellular space in the muscle). This is likely the predominant mechanism of muscle swelling lasting longer than 48 hours.

The upshot of both of these mechanisms is that post-exercise muscle swelling should scale with the amount of stress – both energetic and mechanical – that a muscle experiences during a workout.

All else being equal (i.e., if you control range of motion, proximity to failure, etc.), higher training volumes increase the amount of stress placed on a muscle. So, higher training volumes should lead to greater disruption of muscle fiber membranes, more oxidative stress, more muscle damage in general, a larger inflammatory response, and ultimately, greater muscle swelling.

Furthermore, the time course of muscle swelling is important. While certain markers of muscle damage (acute strength decrements, soreness, and some inflammatory biomarkers) peak 1-3 days after a training session, muscle swelling can increase for several days, with some studies finding that it peaks between around 4-10 days after a workout.

However, most studies that assess hypertrophy following chronic training interventions perform post-training assessments of muscle thickness or cross-sectional area a mere 48-72 hours after the final training session. At this time (it’s argued), significant muscle swelling is still present. So, these studies aren’t just measuring hypertrophy – they’re also measuring the muscle swelling that’s still present after the final training session. As a result, training methods that cause more muscle swelling will also appear to cause more hypertrophy. But, this apparent effect is just an artifact of the muscle swelling.

So, higher training volumes increase the stress on the muscle, which will increase the degree of muscle swelling that occurs. Post-training hypertrophy assessments are performed during a time window where significant muscle swelling would still be present. Since higher-volume training should cause more swelling than lower-volume training, any apparent “hypertrophy” differences between higher- and lower-volume training are probably primarily attributable to differences in swelling, rather than true differences in hypertrophy.

The case against swelling

The case for swelling seems fairly solid until you notice one very important fact: it’s almost entirely reliant on studies on untrained lifters, or subjects performing exercises they’re unaccustomed to. Training experience and exercise familiarity radically change the picture.

When you perform a type or amount of exercise that you’re unaccustomed to, you experience significant muscle damage and significant muscle swelling. However, following this initial bout, your muscles experience a collection of adaptations that result in the “repeated bout effect” (RBE). The RBE refers to the dramatic reduction in muscle damage that occurs as you become accustomed to exercise. With this dramatic reduction in muscle damage comes a dramatic reduction in muscle swelling.

A 2022 study does a good job of illustrating this effect. In this study, muscle damage was induced with a downhill walking protocol. Downhill walking is commonly used in muscle damage research because it produces a lot of muscle damage (resulting from thousands of eccentric quadriceps contractions), it carries minimal risk, and it doesn’t require any training experience to perform properly.

Subjects completed two downhill walking sessions while carrying a load equal to 30% of their body weight. Each session was 45 minutes long, and involved walking down a 25° decline at a speed of 4.5 kilometers per hour (about 2.8 miles per hour). The sessions were separated by two weeks.

Following the first session, subjects got quite sore (reaching a soreness of 6 on a 0-10 scale), and experienced a large increase in creatine kinase levels (a biomarker of muscle damage). Creatine kinase levels peaked at 24 hours following the training session, while soreness peaked at 48 hours post-exercise. And, as expected, the subjects experienced considerable muscle swelling: rectus femoris and vastus lateralis thickness both increased by around 10%.

However, following the second training session, soreness dropped by two-thirds (peaking at a 2 out of 10), creatine kinase elevations were dramatically reduced, and the subjects experienced essentially no muscle swelling.

If you’re skeptical that research on downhill walking won’t directly translate to resistance exercise, similar effects have also been observed with biceps curls.

In a 2015 study, untrained subjects completed two training sessions, each consisting of 10 sets of 6 maximal eccentric curls. The two sessions were separated by 4 weeks.

If you’ve never had the pleasure (displeasure?) of doing maximal eccentrics on a dynamometer, it’s hard to express just how much more brutal they are than standard eccentric/concentric reps you’d perform in the gym. You might perform sets of 6 with around 80-85% of your concentric 1RM. But, you can produce more force eccentrically than concentrically, and a dynamometer ensures that every rep is maximally challenging. So, every rep is performed with your moment-in-time eccentric 1RM.

To illustrate the brutality of this protocol, subjects tested their maximum isometric contraction strength before and after the training protocol. Post-exercise, their ability to generate force had decreased by nearly 50%.

Much like we saw in the downhill walking study, the first training session caused a prolonged decrease in strength (which still wasn’t fully recovered 7 days after the first training session), quite a lot of soreness, a large increase in creatine kinase, considerable muscle swelling, and a large increase in ultrasound echo intensity (which is an indirect marker of increased fluid content in the muscle).

However, following the second training session, all markers of muscle damage were dramatically attenuated.

And, most importantly for our purposes here, the maximal post-exercise increase in muscle thickness was reduced by around 75% (from 4mm to just 1mm), with no muscle swelling present at 48-72 hours post-training.

Furthermore, it’s hard to say that this minuscule increase in muscle thickness was actually due to swelling, since ultrasound echo intensity didn’t increase at all following the second training session.

However, you might still be skeptical. In both of these studies, all of the subjects completed identical workouts. If we want to be even more confident that swelling doesn’t explain the apparent hypertrophy differences between lower-volume and higher-volume training (or less vs. more stressful/challenging training in general), it would be nice to see research showing that more challenging training is more effective than less challenging training for promoting the adaptations that protect the muscles against exercise-induced muscle damage and swelling. Otherwise, you might assume that the relative difference in swelling caused by more vs. less challenging training would be preserved over time, even if the absolute degree of swelling decreased. For example, if low-volume and high-volume training cause muscles to swell by 5% and 10% respectively in untrained lifters, you might expect that relative difference to be maintained, even as the absolute degree of swelling decreased. After a few workouts, maybe low-volume training would cause muscles to swell by 2%, vs. 4% for high-volume training.

So, let’s turn our attention to a study by Chen and colleagues. In this study, subjects in four groups performed 30 eccentric biceps curl reps with loads equal to 40%, 60%, 80%, or 100% of their maximum isometric force. Researchers monitored strength recovery, blood markers of muscle damage, and changes in arm circumference for 5 days following this workout. Then, 2-3 weeks later (once maximum isometric force had returned to baseline levels), the subjects completed a second workout; in this workout, subjects in all groups performed 30 eccentric reps with 100% of their maximum isometric force.

Across all measures, it was clear that the protective adaptations against subsequent muscle damage scaled with the difficulty of the first workout. The subjects who only trained with 40% loads in the first workout experienced larger decreases in strength, slower strength recovery, more swelling, more soreness, and larger elevations in blood levels creatine kinase and myoglobin following the second workout than the subjects who trained in 100% loads in both workouts. In fact, the subjects who trained with 100% loads for both workouts experienced a bit less muscle swelling following the second workout than the subjects who trained with 40% loads experienced after the first workout (where they just used a very light load).

The most intense training session reduced markers of muscle damage by 65-100% in the second training session, and there was a clear linear trend across the four groups for all measures: the more challenging your first workout, the smaller the increases in markers of muscle damage in the second workout.

But, you may still be skeptical. In all of these studies, subjects only performed two training sessions separated by at least two weeks. Perhaps this gave the subjects plenty of time to adapt to the initial stressor, but results would be different with consistent weekly training. If the subjects instead had to train again before their muscles were fully recovered, perhaps this would interfere with these protective adaptations.

With that in mind, let’s check out a 2016 study by Damas and colleagues. In this study, untrained subjects completed 10 weeks of lower body training consisting of 12 sets of quad training per week, split into two sessions. In each session, subjects performed 3 sets of leg press and 3 sets of knee extensions to failure with a 9-12 RM load (loads were adjusted up or down if a subject completed more than 12 reps, or fewer than 9 reps in a set).

Before the first training session, at the start of week 3, and at the start of week 10, the researchers assessed vastus lateralis cross-sectional area, vastus lateralis echo intensity (a proxy for muscle swelling), and blood markers of muscle damage (myoglobin) and inflammation (IL-6). These measurements were taken 72 hours following the previous training session.

Following two weeks of training (4 training sessions), elevations in echo intensity, myoglobin, and IL-6 suggested that the subjects were experiencing considerable muscle damage, inflammation, and swelling. However, by the start of week 10, echo intensity scaled to cross-sectional area, myoglobin, and IL-6 were all back to baseline levels. So, despite the fact that subjects were clearly training again before they were fully recovered during the first few weeks of the program, results suggest that their muscles were still able to adapt, such that no detectable muscle damage or swelling was present at 72 hours post-workout by the end of the training program.

But, you might still be skeptical. After all, that is just a single study, and CSA-adjusted echo intensity is only a proxy for swelling. So, it might be nice to see another study on the topic – and preferably one that directly measures the time course muscle swelling after training.

So, let’s check out a study by Farup and colleagues. In this study, untrained subjects completed 6 weeks of low-load biceps curl training using a within-subject unilateral design. Both arms trained 3 times per week, and performed 4 sets of biceps curls to failure with 40% of 1RM in each session (for a total weekly volume of 12 sets per arm). One arm trained with blood flow restriction, and the other arm didn’t use blood flow restriction.

The researchers assessed muscle swelling for up to 48 hours post-exercise by measuring biceps muscle thickness via ultrasound. These measurements were taken at the start of the study (during a habituation bout), and at the start of week 6. The researchers also monitored self-reported soreness throughout the training program.

Following the habituation bout, some muscle swelling was still present at 48 hours post-exercise (primarily in the blood flow restriction condition). However, by the start of week 6, muscle size was fully back to baseline at 48 hours post-exercise. Furthermore, this decrease in muscle swelling was accompanied by an almost complete lack of soreness after the first ~3 workouts.

But, you might still be skeptical. After all, these studies by Damas and Farup suggest that subjects can adapt to a reasonable (though still challenging) dose of training in 6-10 weeks. But, maybe we would still see a ton of muscle damage lingering around if subjects were instead subjected to a much more brutal training protocol.

So now, let’s turn our attention to a 2021 study by Margaritelis and colleagues. In this study, two groups of untrained subjects completed 10 weeks of training, with one quad training session per week. One group completed 5 sets of 15 maximal concentric reps in each session, and another completed 5 sets of 15 maximal eccentric reps. As mentioned previously, maximal eccentric training protocols are particularly sadistic. Following the first training session, subjects in the eccentric group were still about 20% weaker than baseline 5 days later, on average. Other research has found that some untrained subjects require more than a month to recover from a similar protocol. I promise you that the eccentric training protocol used in this study causes dramatically more muscle damage than whatever quad training you’re currently doing. And, on the flip side, concentric-only training is known to cause very little muscle damage.

In this study, the researchers monitored markers of muscle damage and recovery every week as the training progressed. Unfortunately, they didn’t assess muscle swelling, but they did assess nearly every other marker of muscle damage: pain-free range of motion, muscle soreness, recovery of eccentric, concentric, and isometric force, creatine kinase (as a marker of muscle damage), and C-reactive protein (as a marker for inflammation).

In week 1, the eccentric training group experienced larger indications of muscle damage, fatigue, and inflammation than the concentric training group across literally every measure. But from there, the subjects in the eccentric training group progressively adapted to the enormous stressor they were subjected to. By week 10, there were no longer any significant differences between the eccentric and concentric training groups (in terms of post-workout indicators of muscle damage and recovery), and in isolation, all measures suggest that the eccentric training protocol was no longer causing any meaningful degree of lingering fatigue, muscle damage, or inflammation whatsoever. In all of the figures below, the white circles show the effects of concentric training, and the black circles show the effects of eccentric training.

As a final note about this study, it’s the clearest evidence we have for the progressive reductions in muscle damage that occur with repeated exposures training (even fairly sadistic training), but since this section of the article focuses on post-workout muscle swelling, it’s worth acknowledging again that muscle swelling is one of the few things this study didn’t assess. However, I should note that it’s extremely unlikely that significant swelling would still be occurring despite the fact that all other indicators of muscle damage (reduced strength, increased soreness, increased blood markers of muscle damage, and reductions in pain-free range of motions) were completely ameliorated. Strength reductions don’t scale 1:1 with inflammation, inflammation doesn’t scale 1:1 with blood biomarkers of muscle damage, biomarkers of muscle damage don’t scale 1:1 with soreness, soreness doesn’t scale 1:1 with swelling, etc., but, all of these indicators are associated. So, while this study didn’t assess post-workout muscle swelling, it’s extremely unlikely that this training protocol was still causing large, long-lasting muscle swelling responses in the absence of all other indicators of fatigue, inflammation, and muscle damage.

But, you might still be skeptical. After all, these are still studies in untrained subjects. Maybe they’re just weak, so they’re not capable of causing as much fatigue and muscle damage as trained lifters when subjected to these protocols. Furthermore, most of these studies have used eccentric-only training, so perhaps the results don’t generalize to “normal” exercises that contain both an eccentric and concentric component.

So, let’s turn our attention to a 2023 study by Trindade and colleagues. This study recruited trained subjects: to qualify, subjects needed at least 3 years of training experience, and they needed to have a bench press 1RM of at least 1.2-times body weight.

This study essentially compared the acute responses to normal bench press reps vs. paused reps. Across two sessions (separated by a week, and performed in a randomized order), subjects completed 10 sets of normal bench press to failure with 70% of 1RM, and 10 sets of paused bench press to failure with 50% of 1RM, with 3 minutes between sets.

The researchers assessed muscle thicknesses and ultrasound echo intensity for the front delts, triceps, and two regions of the pecs pre-exercise, immediately post-exercise, and at 24, 48, and 72 hours post-exercise. For both protocols, muscle thicknesses were elevated at 24 hours post-workout, but by 48 and (especially) 72 hours post-workout, they were essentially back to baseline. I extracted the data using webplotdigitizer to get precise values: 48 hours post-exercise, muscle thickness was increased by about 0.7mm, and 72 hours post-exercise, muscle thickness was increased by about 0.2mm (to be clear, that’s 7/10ths and 2/10ths of a millimeter, not a centimeter). 

Furthermore, ultrasound echo intensity was essentially back to baseline by 24 hours post-exercise.

For what it’s worth, I strongly suspect that these small elevations in muscle thickness at 48 and 72 hours post-exercise would be further attenuated if subjects did 10 sets of bench press to failure twice per week for the next 8-12 weeks. The subjects only reported doing an average of 8.6 sets of bench press per week before enrolling in the study, and research suggests that most people don’t train particularly close to failure when left to their own devices. So, I’m extremely confident that these bench press protocols were considerably more challenging (more volume, and closer to failure) than most of the subjects were accustomed to. And, as we’ve seen in this section of the article, muscle damage and swelling responses decrease over time as lifters adapt to increased training demands.

But, you might still be skeptical. This is supposed to be an article about training volume, and the studies discussed in this section mostly come from other bodies of literature. So, let’s return to the volume literature, and see whether longitudinal hypertrophy outcomes offer us any clues.

Whether or not you believe that some swelling is still present at 48-72 hours post-training (even after subjects have had time to habituate and adapt to a training protocol), I hope we can all agree by this point that post-workout swelling does decrease over time, at least to some degree. And, if that’s the case, it naturally follows that post-workout swelling should have less of an impact on apparent hypertrophy outcomes as training duration increases. In shorter studies, more swelling should still be present, and there’s not as much time for a lot of “real” hypertrophy to occur, so it’s more likely that swelling could be masking the “true” effect (or lack of effect) of training volume on “real” muscle growth. But, in longer studies, there’s more time for “real” hypertrophy to occur, and post-exercise swelling should be further attenuated, meaning that swelling should have less impact on the apparent relationship between volume and hypertrophy.  

In other words, if we see a larger hypertrophy difference between higher and lower training volumes in shorter studies than longer studies, that would suggest that the apparent impact of volume on muscle growth may primarily be due to swelling. Conversely, if we see a larger hypertrophy difference between higher and lower training volumes in longer studies than shorter studies, that would suggest that the apparent impact of volume on muscle growth is primarily due to actual differences in “real” hypertrophy.

So, what do we see? As luck would have it, the interaction plots in the Pelland meta-regression already have us covered. They found a much stronger relationship between volume and hypertrophy in longer studies than shorter studies. This provides pretty strong indirect evidence that the apparent effect of volume on hypertrophy is primarily driven by “true” hypertrophy differences, rather than differences in swelling (unless you wanted to hypothesize that, contrary to all research on the topic, swelling actually increases over time with higher training volumes).

Furthermore, we have direct evidence suggesting that, at least with moderate training volumes (between 10 and 17 sets per week, split between two training session), there’s no swelling still present in trained lifters 72 hour post-training. A 2023 study by Refalo and colleagues examined the effects of training to failure vs. training with 1-2 reps in reserve using a within-subject design. It found that both approaches led to similar hypertrophy, but the training and testing protocol are the most relevant details for our purposes here.

Regarding the training protocol, subjects spent the first four weeks of the study training with their habitual levels of quadriceps training volume. In other words, if a subject was doing 12 sets of quad training per week before enrolling in the study, the researchers had them perform 12 sets of quad training per week for the first 4 weeks of the study. Then, for the last 4 weeks of the study, training volume increased by 20% for all subjects. So, by the time that post-training assessments were performed, all of the subjects only had four weeks to adapt to this increase in training stress.

Regarding the testing protocol, the researchers performed two sets of ultrasound scans for each subject both pre- and post-training to assess the reliability of their hypertrophy assessments. The first post-training scans were performed “at least 72 hours” after the last training session, but the lead researcher confirmed via email that “a majority were 72 hours … there were no more than a few that would have been a little longer (max 96 hrs).” The second set of scans were performed 48-72 hours after the first set (so, 120-168 hours after the final training session).

They found that vastus lateralis and rectus femoris thicknesses were actually trivially (<1%) lower 72-96 hours post-training than 120-168 hours post-training (email correspondence):

In other words, unless you’d like to propose that trained lifters experience a delayed, extremely trivial muscle swelling that doesn’t begin until 5-7 days after a workout, this data suggests that the swelling response had fully resolved in less than 72 hours (or less than 96 hours if you’re being extremely skeptical). And to reiterate, subjects weren’t training with extremely high volumes, but they did perform between 5 and 9 sets of high-effort quad training in their final training session before these scans were performed, and they had increased their training volumes by 20% just 4 weeks prior.

So, let’s briefly recap this section before moving on:

1. When we look at research measuring muscle swelling following unaccustomed exercise, particularly in untrained lifters performing fairly brutal eccentric training protocols, we do see dramatic muscle swelling, which often peaks 4-10 days following a workout. However…
2. Both the magnitude and duration of post-exercise muscle damage and swelling are dramatically reduced following a single prior exposure to a stimulus.
3. This reduction in swelling and muscle damage is progressive. We see the largest decreases between session 1 and session 2, but further reductions occur with additional repeated exposures to the same stressor. After 6-10 weeks, we see that even initially untrained lifters are experiencing no discernable muscle swelling at 48 hours post-workout.
4. This reduction in swelling and muscle damage scales with the magnitude of the stressor. So, even though larger exercise stressors initially cause more muscle damage and more swelling, they also cause greater protective adaptations to help reduce subsequent muscle damage and swelling.
5. It only takes about 8 weeks for untrained subjects to fully adapt to particularly brutal eccentric training protocols, such that there are no longer any discernible indications of post-workout muscle damage.
6. The impact of training volume on hypertrophy is more clearly seen in longer-duration studies (where swelling would be expected to have less of an impact on hypertrophy measurements) than shorter-duration studies. If the apparent impact of volume on hypertrophy was actually driven by swelling, we’d expect to see the opposite.

As a general note before moving on, this section about swelling could just as easily be used to push back against a different (but related) argument against high-volume training: high training volumes cause too much fatigue and take too long to recover from, and as a result, this long-lasting fatigue reduces the mechanical tension your muscles can generate in subsequent workouts, thus reducing the hypertrophy stimulus that can be achieved in subsequent workouts.

However, much like the swelling argument, this argument is also primarily supported by studies finding that higher training volumes result in more fatigue and longer recovery times following a single exposure to a stressor. However, just like post-workout swelling responses, the magnitude of post-workout fatigue and the duration of time required for recovery both decrease rapidly with repeated exposures to the same stressor, as we can see in the studies cited above. In other words, a high-volume workout will be extremely fatiguing the first time you train with high volumes, but after a few weeks of training with high volumes, fatigue and recovery concerns are dramatically reduced.

What if there’s still some swelling?

To be entirely clear before starting this section, I don’t think post-exercise muscle swelling is meaningfully affecting the volume literature: I think the observed positive relationship between training volume and measurements of muscle hypertrophy is truly reflective of increased “real” muscle growth when training with higher volumes.

However, I also wouldn’t be surprised if there’s still (at least occasionally) some muscle swelling present at the point in time when hypertrophy measurements are typically taken (48-72 hours after the last workout). If it is still present, I’m extremely confident that it’s fairly small in magnitude – certainly much smaller and shorter in duration than we observe when untrained lifters perform a large volume of unaccustomed eccentric exercise. But, I don’t think we can rule it out entirely.

So, let’s try to roughly model how the observed relationship between training volume and muscle growth may be affected by residual swelling that’s still present at 48-72 hours following a workout.

As our point of reference, let’s return to the study by Trindade discussed previously, and work with the assumption that it’s fairly indicative of the typical degree of muscle swelling that’s still present at 48-72 hours post-training following a high-volume workout in trained lifters, even after they have plenty of time to adapt to the stressor.³ Where would that leave us?

When averaging the increases in muscle thicknesses still present at 48 and 72 hours post-exercise in this study, the subjects’ muscle thicknesses were still elevated by about 2.5% above baseline (about 4% at 48 hours, and about 1% at 72 hours). 

Most of the volume studies involve training each muscle either 2 or 3 times per week. This study observed the swelling that was still present following 10 sets of bench press in a single training session. So, with 10 sets per workout, and 2 or 3 chest workouts per week, we can take this to be representative of the degree of swelling that would still be present with a weekly volume of around 20-30 sets per muscle group. Let’s split the difference and call it 25.

I think we can safely assume that there’s no swelling present after a volume of 0 sets. And, if we’re assuming that swelling is primarily affecting hypertrophy measurements in high-volume studies, that necessarily means we’re assuming that the swelling still present at 48-72 hours post-exercise is a product of training volume (such that, as volume increases, the amount of residual swelling should increase as well). Since we don’t have a third datapoint from this study to help determine the precise relationship between volume and swelling, it would be most justifiable to assume that there’s a roughly linear relationship:

So, the net effect is that we’re assuming that swelling “inflates” hypertrophy measurements by 0.1% per set. To illustrate, if you’re doing 10 sets per week (around 3-5 sets per muscle group per workout), and your muscles grow by 5%, about 1% of that apparent increase is due to residual swelling that’s still present at 48-72 hours after your final training session (meaning your muscles actually only grew by 4%, after accounting for the residual swelling). If you were instead doing 20 sets per week, about 2% of the increase would be due to swelling.

When we apply this “correction,” this is the net effect:

Ultimately, the magnitude of the effect is reduced, but the overall takeaways are similar: we still see hypertrophy increasing as a function of volume.

Now, you could tweak the model parameters to produce the effect that I think the low volume crowd wants to see. But to do so, you’d need to assume that the effect of swelling is nearly twice as large (0.2% per set, instead of 0.1% per set). For example, in the Trindade study, this would mean that muscle size would need to be increased by an average of 5% instead of 2.5% at 48-72 hours post-exercise. This would roughly mean ~8% swelling at 48 hours post-workout (instead of ~4%), and ~2% swelling at 72 hours post-workout (instead of ~1%). And, quite frankly, that would be extremely difficult to justify – it’s far more likely that the swelling observed at 48-72 hours post-exercise in that study would decrease if the subjects benched with that level of volume for a few months.

In other words, you could massage some assumptions about swelling to produce the graph below, but doing so would require making some assumptions about the impact of volume on muscle swelling that are empirically unjustifiable.

In fact, it would be far more justifiable to assume that post-workout muscle swelling may have some impact on hypertrophy measures, but that this impact would be minimally impacted by training volume. 

The last time that all of this was litigated, James Kreiger pulled together a list of studies that assessed muscle swelling at 48 hours post-exercise in subjects performing unaccustomed exercise. Here are the single-workout training volumes and 48-hour swelling responses in those studies:

If a clear positive relationship between training volume and muscle swelling at 48 hours post-exercise doesn’t jump off the screen at you, there’s a reason for that: it’s not there. Here’s how the results look when we graph them:

Now, to be clear, I don’t take this as strong evidence that there’s no relationship between training volume and muscle swelling. We’re ultimately looking at a fairly small batch of heterogeneous studies performed in different populations, using different exercises, assessing different muscle groups, etc. I’m sure that, in a vacuum, 20 sets would cause more post-workout swelling than 1 set. However, since most volume studies use a weekly frequency of 2 or 3 sessions per muscle group per week, these studies do cover a range that would roughly correspond to weekly volumes of about 8-27 sets, and they don’t suggest that swelling has a strong positive relationship with training volume (which is what you’d need if you wanted to conclude that swelling explains the apparent impact of training volume on hypertrophy) within that range, even in the context of unaccustomed exercise.

Furthermore, as discussed previously, the adaptations that protect your muscles from subsequent damage scale with the magnitude of the training stressor that’s applied. So, you’d likely expect that higher training volumes would cause greater swelling in the context of unaccustomed exercise, but you should also expect that higher training volumes would subsequently result in adaptations conferring greater protection to subsequent muscle damage and swelling. As a result, the gap in the damage and swelling responses to lower and higher training volumes should shrink over time.

So, here’s where I personally land on this topic:

For starters, I would not be at all surprised if there’s still some muscle swelling present 48 hours post-exercise, even in the context of longitudinal training studies where subjects have plenty of time to habituate to their training program. However, after about 4-6 weeks, I’m highly skeptical that there’s any meaningful degree of persistent swelling at 72 hours post-exercise. I personally think that a convention of assessing hypertrophy 72-96 hours after the last workout would probably be a bit better than 48-72 hours.

However, I’m also extremely skeptical that training volume has a large enough impact on persistent swelling to meaningfully impact hypertrophy assessments following a typical 2-3 month training study, even when hypertrophy is being assessed 48-72 hours after the last training session. We just don’t see much (if any) indication of significant persistent muscle damage, inflammation, soreness, etc. after about 3-8 weeks of exposure to a particular training stimulus (around 3 weeks for lower-volume “normal” training, and around 8 weeks for fairly high-volume maximal eccentric training). I can’t confidently say that it has absolutely zero impact until we do have direct research on the topic, but I really struggle to see any way to honestly justify a large anticipated effect.

If there is persistent swelling at 48-72 hours post-exercise, even after lifters have 2-3 months to habituate to a particular training stressor, I’d expect the swelling to be small in magnitude, and only weakly impacted by training volume. Instead of the 0.1% per set modeled directly from the Trindade study, I’d probably anticipate something closer to 0.05% per set.

What about sarcoplasmic hypertrophy?

I mostly wanted to address muscle swelling because I see it invoked more frequently in discussions of training volume, but muscle swelling and sarcoplasmic hypertrophy are sometimes bundled together as joint explanations for the apparent increase in hypertrophy with higher training volumes, despite the (apparent) plateau in strength gains with higher training volumes. However, as we’ve already seen, higher volumes do increase strength gains, so much like muscle swelling, we don’t really need to discuss sarcoplasmic hypertrophy here, since it’s invoked to explain a conflict that’s already been resolved. But, for the sake of thoroughness, I’ll also discuss it a bit.

A key difference between muscle swelling and sarcoplasmic hypertrophy is that we know quite a bit about muscle swelling, but we ultimately know very little about sarcoplasmic hypertrophy. It’s not a topic that’s received much research attention because it’s very tricky to study, and very few labs have the equipment that’s necessary for studying it. But, basically everything we do know is summarized in this excellent paper by Roberts and colleagues.

For starters, its mere existence is still somewhat controversial. I’m on record stating that I believe it exists (one, two), but I’m a bit less confident in it than I was a few years ago. This decrease in confidence in primarily due to two reasons:

1. I’ve gained a greater appreciation for sampling variance.
2. The strongest indirect evidence in favor of sarcoplasmic hypertrophy has been largely refuted.

I’ll start with the second point, since it’s a bit easier to explain. A 2015 study by Meijer and colleagues analyzed the muscle fiber contractile characteristics of bodybuilders, power athletes, and control subjects. They found that the bodybuilders had the largest muscle fibers, but their muscle fibers generated way less force per unit of cross-sectional area than the power athletes and the control subjects. There were other potential explanations for this finding at the time (which I discussed in this article), but the most likely explanation appeared to be that the bodybuilders had experienced significant sarcoplasmic hypertrophy, and thus had a lower density of contractile proteins in their muscles.

This wasn’t direct evidence in favor of sarcoplasmic hypertrophy, but it certainly looked like pretty strong indirect evidence. In studies that directly assessed sarcoplasmic hypertrophy, the observed decrease in the myofibrillar protein fraction of the fibers was typically pretty small. For reasons I’ll explain in a moment, there’s plenty of reason to be skeptical of relatively small observed changes in muscle fiber characteristics. But, in the Meijer paper, the bodybuilders’ fibers produced 41% less tension per unit of cross-sectional area than the control subjects, and 62% less tension per unit of cross-sectional area than the power athletes. These differences implied a very large effect – way more sarcoplasmic hypertrophy than you could reasonably attribute to random chance.

However, a 2021 study by Monti and colleagues provided a very strong reason to be skeptical of the results of the Meijer study. Basically, to assess single-fiber contractile force, you have to “chemically skin” the fiber, so that the contractile proteins can come in sufficient contact with the “activating solution” that forces the fiber to maximally contract. Monti and colleagues found the bodybuilders’ muscle fibers produced just as much tension per unit of cross-sectional area as control subjects when fiber size was measured before permeabilization. However, the bodybuilders’ fibers experienced way more swelling as a result of the permeabilization procedure, which would make it appear as if they produced less force per unit of cross-sectional area. Furthermore, the researchers directly assessed myosin concentrations in the bodybuilders’ fibers, and found that it wasn’t different from the control subjects. Their conclusion: “The results show that high degree of muscle hypertrophy is not detrimental for force generation capacity, as increases in fibre size and force are strictly proportional once the differential swelling response is accounted for.”

So, the Meijer study is out. It appears to have gotten erroneous results due to methodological shortcomings that people weren’t aware of at the time (myself included).

The other reason for my increased skepticism, as mentioned above, is that I’ve gained an increased appreciation of sampling variance over the years.

Most of the studies assessing sarcoplasmic hypertrophy do so via biopsies of the vastus lateralis. The total mass of the quadriceps is around 2kg. A typical biopsy removes about 100mg of muscle tissue. So, a biopsy is ultimately sampling around 1/20,000th of the tissue of the quadriceps. If you measure sarcoplasmic hypertrophy using biochemical analyses, that’s the ratio you’re dealing with. If you measure sarcoplasmic hypertrophy via electron microscopy, you’re generally only going to use about 1-2mg of that sample, which means you’re assessing around 1-2/1,000,000ths of the total tissue of the quadriceps. If you’re inferring sarcoplasmic hypertrophy from single-fiber contractile characteristics, you’ll probably be assessing around 6-12 fibers per subject, out of the ~2.5 million total fibers in the quadriceps (also a very small ratio).

So, when a study observes “sarcoplasmic hypertrophy” (a relative decrease in the contractile protein fraction of the muscle), it very well may truly be observing sarcoplasmic hypertrophy. However, it’s also extremely possible it’s just picking up on a bit of random noise: the pre-training and post-training biopsies just happened to snatch small bits of muscle tissue that randomly had slightly higher or lower concentrations of contractile proteins. In other words, even if sarcoplamic hypertrophy doesn’t actually exist, we’d still expect some number of studies to obtain results that look like sarcoplasmic hypertrophy purely by chance, and the risk of this type I error due to sampling variance is probably greater for sarcoplasmic hypertrophy than for most other outcomes in the resistance training literature.

With all of that said, I do still generally believe in the concept of sarcoplasmic hypertrophy, but as I’ve learned more over the years, I’d say my confidence in the phenomenon has decreased from around 90% to around 70%.

But, the much more salient point is that we simply don’t know that much about the phenomenon if it is real. The Roberts review lays out three hypotheses to explain it, but there’s not rock-solid evidence for any of them:

1. What we think is “sarcoplasmic hypertrophy” is actually just muscle swelling.
2. “Sarcoplasmic hypertrophy” is a relatively transient phenomenon to help set that stage for subsequent myofibrillar hypertrophy.
3. “Sarcoplasmic hypertrophy” is something that muscle fibers do once they’ve gotten too large to continue growing via myofibrillar hypertrophy.

I’ll quickly dispense with options 1 and 3. Option 1 is certainly possible, but I’ve already written everything I need to write about muscle swelling for the purposes of this article (see the section above). As for option 3, the strongest evidence in favor of this option was the Meijer study, so I think it’s on fairly thin ice.

But, option 2 is certainly interesting, and dovetails nicely with a topic I discuss in the FAQs near the end of this article (bioenergetic and transcriptional constraints on hypertrophy). The basic idea is that sarcoplasmic hypertrophy helps facilitate myofibrillar hypertrophy. Essentially, synthesizing and maintaining more contractile proteins is a costly process – there’s a high transcriptional burden on myonuclei, a high translational burden on ribosomes, an energy cost associated with synthesizing those new proteins, and an additional spike in the maximal energy cost associated with muscle contraction once more contractile proteins come online.

So, this model of sarcoplasmic hypertrophy proposes that the purpose of sarcoplasmic hypertrophy is to make sure the fiber is prepared for the impending costs associated with myofibrillar hypertrophy. Before building a ton of new contractile proteins (which could put a squeeze on its limited transcriptional, translational, and bioenergetic resources), it first synthesizes more ribosomes to increase its translational capacity (i.e., the rate at which it can synthesize new proteins) and pumps out a bunch of new proteins involved in energy metabolism (glycolysis and gluconeogenesis especially). This causes the fiber to increase in size, which stimulates myonuclear accretion, thus increasing the transcriptional capacity of the fiber so that it can pump out more RNA blueprints for contractile proteins. Finally, once the fiber has made the necessary preparations, it will begin synthesizing new contractile proteins which can fill the space created during this transient period of sarcoplasmic hypertrophy.⁴

I’ll admit that I quite like this idea. It parallels concepts we’ve already covered in this article (connective tissue adaptations that serve to allow for and facilitate subsequent hypertrophy), and concepts we’ll cover later (capillary and mitochondrial adaptations that function similarly). The opposite order of events would instead imply that myofibrillar hypertrophy forces all other systems of the muscle fiber to constantly play catch-up – the fiber builds more myofibrillar proteins, and then has to scramble in order to accrue the myonuclei, ribosomes, and glycolytic proteins that are required to support those myofibrillar proteins. I’ll acknowledge that this isn’t the most rigorous assessment, but that just conflicts with my understanding of how biological systems tend to function best. If the fiber is going to need to synthesize more ribosomes and glycolytic proteins and accrue more myonuclei to support new contractile proteins either way, I find it extremely plausible that those things come first so that the fiber can add more contractile proteins without the risk of experiencing any bioenergetic, transcriptional, or translational bottlenecks. Or, at minimum, if those adaptations can occur in either order (myofibrillar growth necessitating sarcoplasmic expansion, or sarcoplasmic expansion facilitating myofibrillar growth), I find it plausible that it would generally be preferable for sarcoplasmic expansion to precede myofibrillar growth, and not the other way around.

So, if this hypothesis is correct, then sarcoplasmic hypertrophy would not be an adaptation that’s separate from myofibrillar hypertrophy. Rather, it would be a harbinger and positive indicator of myofibrillar hypertrophy.

“But,” you might be thinking, “what does any of this have to do with training volume?”

It’s hard to say, really. We don’t have any studies that directly determine if high-volume training causes more sarcoplasmic hypertrophy than low-volume training. We aren’t even totally sure that sarcoplasmic hypertrophy exists in the first place. But, if it does exist, and if high-volume training did cause more sarcoplasmic hypertrophy, it’s also not clear that it would be a bad thing. Sarcoplasmic hypertrophy might just be a transient increase in water weight that’s totally unrelated to “real” myofibrillar hypertrophy, or it might be an indication that high-volume training causes more “real” myofibrillar hypertrophy by better facilitating the adaptations that lay the necessary groundwork for more “real” myofibrillar hypertrophy.

A brief epistemological detour 

At this point in the article, I think we should briefly pause and zoom out a bit.

So far in this article, my primary aim has been to examine what the data tells us about all of the topics addressed so far:

• Does the data suggest that higher training volumes promote larger gains in muscle size?
• Does the data suggest that higher training volumes promote larger gains in strength?
• Does the data suggest that strength capacity is fully determined by the number of parallel actin/myosin crossbridges aligned with a muscle’s line of pull?
• Does the data suggest that significant muscle swelling is likely to be impacting the apparent relationship between volume and hypertrophy?

You also may have noticed that the position I’ve been arguing against is primarily a logical case for hypertrophy being maximized at lower training volumes: it starts with an assumption about how myofibrillar hypertrophy should impact strength gains and an assumption about the determinants of muscular strength capacity, makes inferences about the impact of volume on hypertrophy based on (a simplified understanding of) the relationship between volume and strength, and then posits that muscle swelling could explain the apparent divergence between hypertrophy data and strength data.

On the surface, it may look like a simple difference in research interpretation: two people reading the same study (or studies) and reaching different conclusions. However, I think we’re actually dealing with a far more fundamental difference in epistemological commitments.

When it comes to research interpretation, my primary commitment is to empiricism. For an empiricist, data collected from experimentation, measurement, and observation is the primary source of knowledge. The research interpretation that most closely aligns with the data that’s most directly related to the outcome of interest is the most justifiable interpretation, and your confidence in any belief should scale with the quantity and quality of data supporting it.

Conversely, it appears that most of the low-volume crowd has a primary commitment to rationalism. For a rationalist, the primary sources of knowledge are logic and reason. You start with ideas and assumptions you strongly believe to be true, and gain new insights through the careful application of deductive reasoning.

To be clear, I don’t think anyone involved in this conversation is a pure rationalist or a pure empiricist – that’s why I said that there are differences in primary commitments to rationalism or empiricism, rather than exclusive commitments to rationalism or empiricism. A rationalist will still turn to empirical data to determine their initial premises, and an empiricist still needs to apply logic and reason to develop hypotheses, assess the generalizability of the data, etc.

Furthermore, empirical and rational methods of inquiry often arrive at similar conclusions. And, when this occurs, rationalists will be quick to point out that the empirical data supports the logical conclusion, and empiricists will be quick to point out that logic supports the empirical conclusion. When both are available, most people recognize that being able to support your position with logic and data is preferable to only being able to support your position with logic or data. So, it’s often difficult to determine whether someone’s primary allegiance is to rationalism or empiricism.

However, people will reveal their primary allegiance in instances where rational and empirical methods of inquiry lead to divergent conclusions. Sometimes, a seemingly logical position isn’t supported by the data, or the data appears to support a conclusion that doesn’t seem logical. When this happens, an empiricist will tend to favor the position that’s mostly strongly supported by the data, while a rationalist will tend to favor the position that is most strongly supported by logical reasoning. The empiricist will typically try to explain the divergence by attempting to determine where the logic went astray (Did you have a faulty premise? Were there variables you didn’t account for in your reasoning? Is our mechanistic understanding of this phenomenon incomplete?), while the rationalist will typically try to explain the divergence by attempting to determine where the data went astray (were there issues with study design or data collection that explain why the research didn’t reach the conclusions it should have reached?).

Again, I want to emphasize that empirical and rational modes of thought and inquiry aren’t mutually exclusive. An empiricist may find a logical argument to be more persuasive than the empirical evidence in situations where the logical argument is strong and the empirical evidence is sparse and weak, and a rationalist may find the empirical evidence to be more persuasive than a logical argument in situations where the data is strong and the logical argument is fairly flimsy. But, as more and more data is collected, the rationalist will tend to favor logic over data for far longer than the empiricist will.

I think this more-or-less describes the state of play in the ongoing volume debate. As an empiricist, I believe that the data strongly support the position that higher training volumes generally promote more muscle growth, even at pretty high training volumes (>20 sets per week). The primary argument against this position is primarily a rationalist argument, and the rationalists clearly believe that their logical case against the impact of higher training volumes on muscle growth is sufficiently strong to trump the direct empirical evidence.

So, you wind up at a bit of an impasse. The empiricists aren’t convinced by the strength of the rationalists’ logic, and the rationalists aren’t convinced by the strength of the empiricists’ data. This naturally flummoxes all participants, and results in a frosty dialog that goes nowhere. The empiricists already know the rationalists’ arguments – they just find the data to be more convincing. The rationalists are already aware of the empirical evidence – they just find their logical arguments to be more convincing.

When you find yourself confronted with this scenario, who should you trust?

You should trust the empiricists

In the previous section, I tried to be Fair and Balanced™ when describing rational vs. empirical modes of inquiry. But, for the reasons I’ll discuss below, I strongly believe that empiricism is more likely to lead you to more correct beliefs more often than rationalism will. And, I’m not alone in this belief: it’s literally baked into the structure of science.

As mentioned previously, rationalism does play an important role in scientific inquiry. For starters, experimental research typically begins with a hypothesis. To generate a hypothesis, you need to make predictions about the outcomes you expect. You generate these predictions through some rational process (you make deductive inferences from a first-principles application of mechanisms, you believe some other study would have gotten a different result if it was conducted in a different population or with a slightly different methodology, you have a logical basis of predicting that findings in one population will generalize to another population, etc.). Furthermore, once you collect your data, interpreting your results requires some rationalist thought. Your data tell you the impact of a specific experimental manipulation in a discrete pool of subjects, but you need to apply a bit of logic and reason to determine the degree to which you expect your results to generalize to slightly different interventions in slightly different populations. Finally, the process of creating or refining overarching scientific theories necessarily requires some rationalist reasoning. Any (good) theory will certainly be informed by data, but the process of putting the pieces together in a way that accurately describes several lines of evidence, while simultaneously creating new testable hypotheses, necessarily requires reasoning that goes beyond a simple, direct, and literal interpretation of empirical findings.

However, in all of these cases, empirical data is the ultimate arbiter of truth or falsity. You generate a hypothesis with rationalism, but you confirm or refute your hypothesis based on your empirical findings. You apply rationalist reasoning to predict the generalizability of your findings, but your interpretation will either be confirmed or refuted on the basis of subsequent empirical evidence. Rationalism is involved in the process of generating scientific theories, but theories are only considered to be confirmed if they generate testable predictions that are validated by subsequent empirical research, and they’re considered to be falsified if they generate predictions that are refuted by subsequent empirical research.

Now, that’s obviously a very simplified account of how science works. Plenty of empirical evidence is of relatively low quality, the relative strength of the evidence for or against a particular hypothesis or theory isn’t always clear-cut, sociological phenomena can influence the dominant views of a field to shift faster or (more often) slower than the evidence would suggest that it should, observational and exploratory research doesn’t necessarily need to have a hypothesis or be theory-driven, and there are even non-empirical sciences. But, from a high-level view, what I described above is more-or-less how empirical, experimental sciences are supposed to function. You generate ideas with rationalism, and test them with empiricism.

Of course, circling back to the start of this section, none of that actually establishes that “empiricism is more likely to lead you to more correct beliefs more often than rationalism will.” It tells you that experimental scientists (either tacitly or explicitly) believe that to be the case. But what if they’re all wrong? Why else might it behoove you to generally favor empiricism?

Scaling confidence

Empiricism (at least as it’s practiced in the sciences) allows you to scale the confidence of your beliefs with the strength of the evidence in a way that rationalism doesn’t. When evaluating empirical evidence, you can statistically quantify how certain (or uncertain) you are about the existence of the effect, you can directly quantify the strength, magnitude, and/or consistency of the effect, and you can produce reasonable estimates of how variable the effect may be.

Rationalism, when strictly applied, is an all-or-nothing framework. A train of logic necessarily leads to a particular conclusion. If you believe the logic is sound and the premises are true, you should have 100% confidence in the conclusion. If the logic isn’t sound, or at least one of the premises isn’t true, your confidence in the conclusion suddenly becomes some unquantifiable value below 100%. Ultimately, these are both negative outcomes. Having complete confidence in an idea offers a wonderful feeling of security … unless it turns out to be wrong. I’d much rather be able to take the empirical approach of making small updates to my beliefs about a particular topic as the strength of the evidence shifts, rather than only having the options of complete certainty and unquantifiable uncertainty.

Ultimately, I think that’s one of the reasons why rationalist arguments have a tendency to be quite sticky, even as evidence accumulates against them.

An easier process for updating your beliefs

For an empiricist, the process of changing your mind is smooth and non-threatening. Maybe you initially had 80% confidence that a particular intervention had a small-to-medium positive effect, but over a period of 2-3 years the balance of evidence shifts, such that you’re now 70% confident that it has a trivial-to-nil effect. One day it dawns on you that you’ve changed your mind, but you barely even noticed as it was happening.

However, for a rationalist, your belief holds up until it suddenly collapses. As evidence accumulates against your belief, you may experience it as a threat, and you’re forced to deal with more and more cognitive dissonance until you finally realize that it’s untenable. But, this is a fairly painful process, since it generally involves exchanging the comfort of absolute confidence for the wasteland of unquantifiable uncertainty. The best case scenario is that you discard your prior belief because you were won over by some other rationalist argument. This lets you skip over the period of unquantifiable uncertainty, but there’s still some whiplash that comes from exchanging one iron-clad belief for another that may be incompatible with your prior belief.

For what it’s worth, I think this is one of the reasons why some people get frustrated by their perception that “science is always changing its mind about things.” When your typical exposure to online “scientific content” (really, content that merely presents itself as if it were scientific) primarily focuses on logical argumentation from presumed mechanisms, it can look like scientific consensus is constantly shifting. One argument leading to one set of assumed outcomes becomes popular. Then it’s replaced by another popular argument leading to a different set of assumed outcomes. Then another. Then another. It looks like “science” can’t make up its mind.

In reality, there’s very rarely an abrupt sea change when you focus on the data that actually assesses the outcome of interest – just a gradual accumulation of empirical evidence, leading to gradual shifts in effect estimates and confidence levels. Sometimes those shifts bring a null effect estimate into the realm of “statistical significance” or vice versa (“statistical significance” is another big can of worms, but this is not the time to get into it), but it’s extremely rare for the balance of evidence to dramatically change in a short period of time, and even rarer for it to then quickly shift back in the opposite direction.

Less hindered by incomplete mechanistic understandings

Rationalism relies on deductive reasoning. Deductive reasoning only reliably works in the context of relatively simple systems, or systems that are almost perfectly understood – at least if you want to have any reasonable degree of confidence in your conclusions.

In essence, for strings of deductive logic to hold up, you need to be able to make (and justify) strong categorical statements at every step of the argument. If you can’t, certainty progressively disintegrates. So, to have a high degree of confidence in your conclusions, you need an essentially perfect mechanistic understanding of the phenomenon. You need to be able to say “A necessarily causes B, which necessarily causes C, which necessarily causes D, etc.” If you can make this type of statement, then you can make strong inferences about D based on A.

However, biology fundamentally doesn’t work like that. B may result from A under certain conditions, and B might influence C, but only when Z and Y are also present, and C might interact with D, but only when D isn’t experiencing competitive inhibition from X, etc. You can have a reasonable degree of confidence about the impact of A on B, less confidence about the impact of A and C, and even less confidence about the impact of A on D.

If you’re a rationalist in this situation, you basically have three options:

1. Just pretend like you can still draw a straight line from A and D, and ignore all of those pesky complications. If you go down this road, there’s a very high likelihood you’ll be wrong.
2. Abandon any hope of making strong inferences about D based on A. If you go down this road, the best you could say is something along the lines of “A might have some impact on D.”
3. Spend the next decade trying to fully characterize the interactions between A and B, flesh out C’s interactions with Z and Y, understand the upstream factors that influence Z and Y, and then do the same for D and X. Then, you might understand the mechanistic pathway connecting A to D well enough to make deductive inferences about D from A.

If you’re an empiricist, you have a much more straightforward path: design a study where you manipulate A, and then observe the resulting impact on D. Once a few more lab groups conduct similar studies, you’ll have a pretty good idea about the relationship between A and D.

I realize that this may seem pretty abstract, so let’s turn our attention to a concrete example: drug trials. 

The process of bringing a drug to market starts with the discovery of a new drug. If a drug has been “discovered,” that means researchers have identified a compound that mechanistically impacts some biological pathway involved in health or disease. In other words, that means there’s a logical, rational reason to expect that it should exert positive effects of some sort.

Following drug discovery, you move into preclinical trials. During the preclinical phase, you investigate whether the drug actually leads to the desired physiological impacts when tested in cell cultures or (typically) rodents. Sometimes it does. Most of the time, it doesn’t. There may be issues with drug delivery, there may be excessive side effects, or the researchers may just find that the pathway the drug mechanistically affects doesn’t actually yield the anticipated impact.

Most drugs don’t make it out of preclinical trials (well below 1%). For the ones that do make it out, you then start clinical trials in humans. You start with phase 1, which is mostly about safety (i.e., can people actually use enough of the drug that it might theoretically help before it starts causing harm). Then phases 2 and 3 are mostly about whether the drug actually does what you expect it to in humans, whether it works better than the current standard of care, and whether or not the early safety data translates to longer-term safety.

Of the <1% of “discovered” drugs that make it to clinical trials, about 12% actually get approval to be marketed for human use. All in all, well below 1% of new drugs that get discovered actually make it onto the market, because most fail somewhere in the process of gathering experimental data. And keep in mind, if a drug is considered to be “discovered,” that means it’s already known to mechanistically do something that is believed to be useful and valuable.

So, in the pharmaceutical industry, promising drugs with promising mechanisms fail to pan out >99% of the time. They don’t pan out because experimental data tells us they don’t pan out, despite the fact that they were all mechanistically expected to work. Pharma companies wouldn’t dump (collectively) billions of dollars into research on compounds without a clear mechanistic rationale. Say what you will about big pharma, but I have no doubt that pharmaceutical researchers understand biological mechanisms better than most fitness influencers, but their track record suggests that their logical, rational predictions about the impact a drug should have are essentially wrong >99% of the time.

“But,” you may argue, “we already have a complete mechanistic understanding of hypertrophy! It’s just a matter of exposing muscle fibers to a sufficiently high degree of mechanical tension.”

However, that’s crucially not how you’d give a mechanistic account of a complex phenomenon. That just tells you that you believe a stimulus to be associated with the outcome of interest. A complete mechanistic accounting of hypertrophy would be able to precisely tell you how the phenomenon occurs. Just to illustrate the limits of our current understanding, if we did have a robust mechanistic understanding of hypertrophy, and we knew it was all a matter of mechanical tension, these should all be very easy questions to answer, complete with specific citations, and thorough refutations of all lines of research that would suggest alternate answers:

• Specifically, how much tension is required to stimulate a hypertrophy response?
• How can we even assess per-fiber mechanical tension in vivo during dynamic contractions?
• What is the precise dose-response relationship between fiber tension and post-exercise protein synthesis?
• What is the precise relationship between post-exercise protein synthesis and chronic hypertrophy?
• How, exactly, is mechanical tension sensed by muscle fibers? There are at least 4 or 5 candidate sensors that have been identified, but some of the signaling pathways upstream or downstream of these sensors are poorly characterized, and there’s evidence both for and against most of them. So, which ones are actually critical?
• How can we be confident that we’ve actually identified all of the sensors of mechanical tension?
• How can we be confident that mechanical tension is the sole initiator of these signaling pathways at the exclusion of all other stimuli?
• How can we be confident that tension per se is the only factor influencing signal transduction through these signaling pathways, and other stimuli don’t enhance or dampen these signals prior to the initiation of protein synthesis?
• Most of the mechanistic evidence comes from in vitro experiments, genetically modified rodents, or rodents exposed to extreme protocols to induce overload that don’t closely mimic “normal” resistance training (for example, synergist ablation or unilateral diaphragm denervation). How confident can we be that these mechanisms directly translate to humans performing normal resistance exercise?
• A mechanical tension stimulus is primarily believed to contribute to hypertrophy through mTOR-dependent signaling pathways, but there’s evidence for hypertrophy signaling through non-mTOR-mediated pathways. How can we be sure that the initiation and signal transduction through these pathways is solely due to tension?
• Factors like mitochondrial adaptations, increases in capillary density, ribosome biogenesis, and myonuclear accretion appear to play important roles in mediating or moderating hypertrophy responses. How can we be confident that tension per se is a sufficient stimulus to optimize these adaptations to ensure that a muscle can experience a maximal hypertrophy response?
• Why do fibers associated with low-threshold motor units still experience hypertrophy following purposeful resistance training, when the Size Principle would suggest that they should be experiencing maximal tension many times throughout the day when simply performing activities of daily living?

I could keep going, but I hope you get the point. Currently, the answer to all of these questions is something along the lines of “we don’t yet know (either in part or in full),” or even, “we don’t yet have the experimental models or measurement techniques that would be required to start answering that question.” And notably, would all be very basic questions if we did actually have a robust mechanistic understanding of hypertrophy. Furthermore, all of the partial, tentative, or speculative answers we have to all of these questions just give way to a similar list of additional follow-up questions. But, those are the types of questions we’d need to have clear, definitive answers to in order to claim that we have a strong mechanistic understanding of hypertrophy.

If you’ve made it this far in the article, this is clearly a topic that you’re deeply invested in, and you clearly have the attention span to read long, dense articles. So, I’d recommend that you take some time to read a 2023 review paper by Roberts and colleagues titled “Mechanisms of mechanical overload-induced skeletal muscle hypertrophy: current understanding and future directions.” For my money, it’s the most thorough review we have of the topic (even though it may already be out of date; a new upstream regulator of the mTOR pathway that’s influenced by resistance training was potentially identified just a few months ago). But, if you don’t have the time or desire to read it, I think the conclusion should give you a pretty good vibe about the current state of mechanistic hypertrophy research:

“Skeletal muscle hypertrophy research has rapidly evolved since the landmark report by Morpurgo in 1897. Pioneering discoveries in the field have motivated others to adopt innovative methodologies and drive the research boundaries in meaningful ways. Given the rapid advancements in molecular-based research techniques, investigations in upcoming years will continue to confirm or refute which of the discussed mechanisms are obligatory for (rather than coinciding with) load-induced skeletal muscle hypertrophy. More importantly, these efforts will likely unveil novel mechanisms that continue to reshape our thinking in this area of muscle biology.”

This group of experts – all of whom are leading researchers who are actively investigating the mechanisms of hypertrophy – still have a lot of open questions on the topic. So, anyone who believes that they already have a more-or-less complete mechanistic accounting of hypertrophy is either decades ahead of the science (which would involve knowing things that are currently empirically unknowable given our current methodological limitations), or painfully deluded in a manner that belies extreme arrogance, extreme ignorance, or utter incuriosity. 

The appeal of rationalism

The main reason there’s an entire section of this article about epistemology (and, if I’m being honest, one of the main reasons I’m writing this article in the first place) is that I think rationalism is crushing empiricism in the “evidence-based” fitness community, and I think that’s a bad thing for information consumers. But, I understand the appeal. So, I’d like to briefly explain why (I believe) it’s so compelling, and the incentives that favor content based on rationalism over content based on empiricism.

Empiricism is only exciting if you’re a massive nerd about a particular topic. If you’re deeply invested in the process of learning the intricacies of new experimental methodologies to test tricky hypotheses, or if you enjoy learning about new statistical techniques to more precisely account for latent confounders (etc., etc.; you get the point – nerd shit), then empiricism is a blast. Otherwise, it’s incredibly boring, and empirically-driven content may even come across as incurious, bordering on anti-intellectual.

To gain a thorough empirical understanding of a topic, you need to find all of the research assessing the outcome of interest, read all of the research assessing the outcome of interest, extract a bunch of data from tables and figures, and then statistically analyze it.

No step of this process is particularly exciting.

And, if you’re a content creator, the return on your time investment is minuscule. “I did about 60 hours of work to extract and analyze all of the data on this topic. After all of that, I can tell you that the effect size associated with this particular variable is around d = 0.21, with a 95% confidence interval of d = 0.03-0.39. So, I’m pretty sure it has a trivial-to-small positive effect. Hope you enjoyed this content!”

Furthermore, in a competitive online ecosystem where everyone’s vying for your attention, it’s painfully difficult to stand out and make a name for yourself. The data are what they are, and there are a very finite number of (justifiable) methods of analysis. Once there’s a pretty good meta-analysis or meta-regression on a particular topic, you have your answer about the effect of a particular intervention or variable. Notably, it’s going to be the same answer as every other empiricist on the planet.

Additionally, an empirical epistemic approach lends itself to cautious, guarded explanations and/or extrapolations, which may come across as a lack of expertise or curiosity (in reality, it’s just extremely tricky to empirically establish causation).

For example, instead of saying “A caused B because of C,” you might say, “We observed B after A. A is associated with C, and some rodent research suggests that C might mediate B, so C might be one of the factors contributing to the relationship between A and B.”

In almost all cases, the latter statement is the more epistemically justifiable statement. You can’t say “A caused B because of C” unless a study on A assessed both B and C, and formally established causal links between A and B and between B and C (and even then, unless you’re dealing with direct 1-to-1 causation – and you never are – the best you could say it that C causally explains some percentage of the variance of A’s effect on B). However, the more guarded explanation is almost always worse for content. When you spell out the tenuous connections between A and B and between B and C, it looks like you’re merely speculating instead of providing a solid mechanistic explanation™. Or it may cause people to ask why you’re not confident about this topic when everyone else does seem to be sure – do they just understand it better than you do? This can lead to caution about discussing mechanisms and explanations for empirical results, which can come across as incuriosity or anti-intellectualism.

Finally, empirical arguments and disagreements are almost always very boring, very long, and/or very pedantic (exhibit A: this article). You’re going to be extracting and analyzing a bunch of data, you’ll need to explain the process of doing so (because most people in your audience won’t follow what you’re saying if you don’t), and most of the time, the net result will just be, “the data isn’t strong enough to support the type of statement you’re trying to make.” All the while, you’ll be using language that projects less confidence than whoever you’re disagreeing with. So, if you want to engage in “the discourse,” you’ll probably be wading in two weeks after everyone else has moved on, with a lukewarm take that most people don’t have the attention span to digest. In contrast, it doesn’t take long at all for a rationalist to cook up some argument that sounds reasonable enough, and maybe even find one or two citations that appear to support each point.

Unlike empiricism, rationalism makes for great content that more naturally resonates with people.

To start with, “soft” rationalism is our default way of thinking and making decisions in most contexts that actually require purposeful thought. Empiricism works, but it’s extremely unwieldy. You don’t have time to conduct a meta-analysis or independently collect and analyze a bunch of data to answer most of the questions and address most of the problems you have in life. Instead, you use (typically informal) modes of logic and reason to make decisions, resolve disagreements, etc. When confronted with a well-constructed rationalist account of some phenomenon, it’s more likely to resonate, because it closely reflects the structure of the way you’d likely think through the issue on your own.

I think rationalist arguments are also more memorable than empirical results. Remembering a well-constructed argument is a bit like remembering a story or narrative, where there are multiple pieces of associated information. If you can remember part of it, those associations help you recall the rest. You don’t necessarily need to remember specific values or citations – just how the pieces of the story relate. On the flip side, remembering the numerical results of a meta-analysis is, pretty plainly, just a matter of memorizing a discrete fact, which will be much more likely to slip your mind. 

As mentioned above, the presentation of a logical argument also more closely conforms with most peoples’ understanding of expertise. It gives you the impression that you’ve thought deeply about the topic, put all the pieces together, and confidently arrived at a new insight through the strength of your powers of reason. In making the argument, you’ve provided the listener with numerous (reasonable-sounding) explanations that appear to support your conclusion. And, as a result, the person you share it with will be likely to share it with other people, because doing so will give them a great opportunity to look smart and demonstrate their (presumed) understanding of the topic. In contrast, presenting an empirical case mostly goes something like, “so-and-so collected the results from all of the studies on a topic, did a meta-analysis, and this was the result.” If you’re talking to an empiricist, this will go over great. If they have a follow-up question, it might be something like, “do you remember the inclusion and exclusion criteria?” But, if you’re talking to most people, it’ll sound like you’re just parroting a decontextualized figure someone else came up with. Their most likely follow-up question will be to inquire about an explanation for the results. A response of “I don’t know for sure” will sound like incuriosity, and a response of “I don’t know for sure, but here’s a list of all of the proposed explanations, and the strengths and weaknesses of the data supporting each one” will put them to sleep.

For these reasons, content with a rationalist flavor is more “viral” than more strictly empirical content:

1. It’s easier and quicker to make.
2. It has a structure that feels natural, and lends itself to a presentation style that conveys the impression of expertise.
3. It presents arguments that are easier for people to remember and share with others.
4. It’s more amenable for real-time engagement with online discourse.

As a result, when empiricists and rationalists disagree, rationalists have the rhetorical upper hand.

It’s not particularly difficult to cast doubt upon empirical results. Here’s a handy list of smokescreens: 1) These studies were too short, 2) the sample sizes were too small, 3) the subjects didn’t have enough prior training experience, 4) the researchers didn’t account for [insert potential confounder here that could only be controlled for if the subjects lived in the lab]. 

You could also cook up boutique criticisms for certain research topics (“The hypertrophy data may have been corrupted by muscle swelling!”), but that handy list should meet most of your anti-empiricism needs (note: feel free to ignore this list when you feel compelled to find citations that support any points in an argument you’d like to construct). And, when all else fails, you can simply do a bit of circular reasoning and use your argument that’s contradicted by the empirical results as evidence against the empirical results (the observed results don’t logically follow from your argument, so the problem must be with the empirical results). You can accomplish all of this in a 2-3 sentence comment or a 30-second video.

However, it’s often quite difficult to negate a (formally valid) rational argument in a way that’s rhetorically effective. On a purely technical level, it’s a breeze. If you just point out that one of their premises isn’t adequately supported, or that a particular conclusion is only related to, associated with, or sometimes resulting from a particular premise (when it would need to necessarily follow from the premise), that should be sufficient to demonstrate that the argument relies on an unsupported assumption, or that the argument is unsound and the conclusion is untenable. But, that’s not rhetorically effective – it’s just going to come across as pedantry. So, you typically only have 1-3 tools in your arsenal.

The first is just to pick some low-hanging fruit. A rationalist is arguing that the empirical results shouldn’t be trusted because they don’t conform to the expectations of their logical argument. You can turn that around and note that the logical argument shouldn’t be trusted because, if it were true, it would be supported by the empirical results.

For this particular article, I didn’t think this would be rhetorically effective, since the argument I’m contending with contained a built-in explanation for why the empirical results didn’t conform to the expectations of the argument (“we don’t see a plateau in hypertrophy measurements at lower training volumes because higher training volumes cause more muscle swelling”).

The second is to demonstrate that a premise of their argument is untrue. This can be a good strategy, but it’s most effective when it’s fairly quick and easy to demonstrate that a premise is untrue. That’s usually not the case, however. An argument is less likely to gain traction if it contains a premise that’s too easy to disprove. If a popular argument contains an untrue premise, it’s usually untrue for a reason that wouldn’t be entirely obvious to most people, and would thus require quite a bit of explanation.

Most of the first part of this article employed this strategy. The argument that higher training volumes only appear to increase hypertrophy because of muscle swelling relied on two disprovable premises: 1) muscle force capacity is fully determined by the number of parallel actin/myosin crossbridges, and 2) higher training volumes don’t lead to larger gains in muscle strength. I think I did an adequate job of disputing those premises, but doing so required around 4,000 words per premise. So, I’m not sure how rhetorically effective it will be (not to mention the difference in time cost between formulating a thorough empirical case and slapping together a reasonable-sounding argument). 

An example of an instance where it is easy to disprove a premise would be if a premise is supported by a single erroneous citation, and you can show that the citation actually negates the premise it’s supposed to support.

The third is the classic reductio ad absurdum: simply demonstrate that the argument would lead to untenable conclusions.

In this article, that was the point of the section, “A simple guide to maximize hypertrophy with zero effort.” When the argument I’m contending with is only applied to the volume literature, it leads to a conclusion that plenty of people find to be quite reasonable: you don’t need to train with high volumes in order to maximize muscle growth. However, by pointing out that the same argument would also apply to the proximity to failure literature, I showed that it leads to a conclusion that is clearly incorrect: you can maximize muscle growth by doing 5 sets per week with 10 reps in reserve.

Of these three options, the reductio ad absurdum is the most rhetorically effective tool, in my opinion. It’s usually quick to deploy, results in an argument that’s easy to understand, and leaves your opponent looking pretty dumb. But, it’s not always an option on the table.

More often than not, you’re stuck with a slow, laborious application of option 2, and that’s still a pretty favorable outcome, all things considered. Often times, you do just have to play the role of pedantic logician, because you’re dealing with an argument where you can’t demonstrate that the premises are false (not because they’re correct, but because there’s simply insufficient research to support or refute them), so you have to fall back on critiquing the mechanics of the argument itself. Typically, that means you just have to point out that it relies on an unsupported assumption, and the rhetorical effectiveness of your argument simply depends on whether anyone actually cares that a reasonable-sounding premise is technically just an assumption. 

How to think, not what to think

I’ll now admit what you may have already been suspecting: this article was another Trojan Horse. Am I really invested in one particular position in the volume debate? Not really. My commitment is to a particular method of inquiry, not to a particular outcome. Right now, I think the evidence supports the position that higher training volumes are better for hypertrophy, but if the evidence shifts, my opinions will shift as well. On a purely selfish level, I’d love for that to happen. I personally don’t have as much time to train as I once did, and I’d love to be convinced that I can still maximize my results with less training volume than I currently believe to be necessary. And, as a business owner, the low-volume position is clearly the more marketable position, so it would behoove me to be able to hold (and market) that position in good faith – people want to hear that they can get better results with a smaller investment of time and effort.

But, first and foremost, I aim to be a responsible science communicator. And ultimately, science is a mode of inquiry, not a prescribed set of beliefs. My beef with most of the low-volume crowd – and more generally, my beef with a lot of the current batch of fitness influencers who adopt the veneer of caring about research – isn’t that I disagree with their discreet beliefs about training. My main problem is that they present their conclusions as if they’re scientific, but they reached those conclusions using a method of inquiry that is fundamentally unscientific, or even antiscientific.

As mentioned above, the scientific process starts with a hypothesis. You design an experiment to test your hypothesis. You collect data from the experiment. You analyze the data, which offers evidence for or against your hypothesis. At the end of this process, the direction and strength of your belief should scale with the direction and strength of the evidence. And, as more and more research is done on the same topic, it will produce more empirical evidence that will help further inform your beliefs.

But, at no point in this process do you just turn your brain off and stop asking questions. You might have reason to believe that different results would be seen in a different population. Or, you might have reason to believe that the results are confounded in some way, and that controlling the confounder would produce different outcomes. Or you might want to investigate the mechanistic processes that explain the outcomes you’re observing.

In all cases, those rational processes allow you to form new hypotheses: they bring you back to step 1 in the process. But, to know whether your hypothesis is correct or incorrect, you need to test it (or you need to wait for someone else to test it, if you’re not a researcher).

Rationalism as a primary mode of inquiry and as a primary epistemic commitment fully inverts this process. Instead of evaluating the veracity of your hypothesis based on the strength of the most directly relevant evidence, you evaluate the strength of the evidence based on whether or not it conforms to your hypothesis. It skips (or simply ignores) the part of the process where you collect data to determine whether or not your hypothesis generates predictions that turn out to be true. It simply assumes that a particular conclusion is true because it’s the logical implication of your hypothesis, regardless of whether it comports or conflicts with the data.

A key component here is that empirical beliefs should be most strongly informed by the most direct evidence, whereas hypotheses are typically generated from forms of indirect evidence. You can see how this plays out when we look back at this article.

We start with a meta-analysis suggesting that higher training volumes lead to more hypertrophy. As far as I’m aware, no one is claiming that the hypertrophy results themselves, as they exist in the current literature, actually support an alternate conclusion. For an empiricist, this is the strongest base of evidence we have: a pretty large body of literature where a single variable (training volume) is manipulated, leading to direct, measurable differences in the outcome of interest (muscle size).

But, someone might look at that body of research and have a completely reasonable, rational thought: higher training volumes are a larger stressor than lower training volumes. Larger stressors tend to cause more muscle damage and swelling. Swelling isn’t always dissipated 48-72 hours after a training session. So, it’s possible that higher training volumes are predominantly causing more swelling, rather than more actual muscle growth.

That’s an entirely reasonable thought to have. But, crucially, this is a hypothesis informed by indirect evidence. As discussed above, I’m personally skeptical that post-exercise swelling is having a particularly large influence on the aggregate results of the volume literature, but it’s a topic for which we don’t yet have direct empirical evidence. I’d love to see research designed to test this hypothesis. But again, this is a hypothesis.

If you find it to be a compelling hypothesis, this would be a perfectly justifiable statement: “The volume research suggests that higher training volumes lead to more muscle growth. However, this research typically performs post-training assessments of muscle size 48-72 hours after the last training session. Some research suggests that muscles may still be impacted by swelling during this time window, and higher training volumes may cause more swelling than lower volumes. So, I personally have some reservations about the applicability of studies that test particularly high training volumes. We’ll know whether these concerns are justified once we have several studies assessing the time course of post-exercise muscle swelling following longitudinal training programs with varying training volumes.”

In contrast, you cannot construe your personal misgivings about the time course of muscle swelling as direct (or even indirect) evidence against the empirically-established relationship between volume and hypertrophy, and you especially can’t construe it as evidence for a belief that hypertrophy is maximized at low training volumes. It’s simply a reason for skepticism that can be validated or assuaged once there’s applicable data to support or refute it.

Similarly, someone might look at the volume/strength meta-regression in the Pelland paper and have a completely reasonable, rational thought: it’s believed that increased muscle mass should contribute to larger strength gains, because increased muscle mass typically results in a larger number of parallel contractile proteins. So, when it looks like higher volumes promote larger gains in muscle size but not larger strength gains, it’s possible that the apparent increases in muscle size at higher volumes are due to sarcoplasmic hypertrophy or muscle swelling, rather than “real” myofibrillar hypertrophy.

Once again, that’s an entirely reasonable thought to have. But, once again, this is a hypothesis informed by indirect evidence.

If you find this to be a compelling hypothesis, this would be a perfectly justifiable statement: “The volume research suggests that higher training volumes lead to more muscle growth. However, research looking at the relationship between volume and strength gains tends to suggest that strength gains plateau at lower training volumes. It’s generally assumed that greater hypertrophy should contribute to larger strength gains, because “real” myofibrillar hypertrophy entails a gain in the number of parallel contractile proteins. So, I personally have some reservations about the research finding that higher training volumes promote more muscle growth, because the current research can’t rule out the possibility that these results are influenced by muscle swelling or sarcoplasmic hypertrophy. But ultimately, more research is needed to clear up this apparent conflict, which will let me know whether my reservations are justified.”

But, once again, you can’t just look at this apparent conflict and conclude that the hypertrophy results must be wrong because the dose-response relationship between volume and hypertrophy appears to differ from the dose-response relationship between volume and strength. You may simply have an incomplete understanding of the mechanistic contributors to whole-muscle strength, the apparent conflict may simply be the result of the two meta-regressions using different sets of studies, or it may even suggest that we have an incorrect understanding of the volume/strength dose-response relationship instead of the volume/hypertrophy dose-response relationship (i.e., “strength gains don’t actually plateau at low training volumes,” is just as parsimonious of an explanation as “hypertrophy does actually plateau at low training volumes”).  

In short, a scientific epistemology doesn’t require a slavish devotion to direct empirical evidence at the exclusion of reason, logic, or indirect evidence. Rather, direct empirical evidence is the primary basis for understanding a phenomenon, and reason, logic, and indirect evidence help provide a deeper understanding of the phenomenon (helping to explain why it occurs when it appears that you have a strong mechanistic rationale for your direct empirical observations), nuance to your understanding of the phenomenon (helping you determine when and for whom the results may or may not generalize), or even reasons to be skeptical of the direct empirical evidence (when logic, reason, and indirect evidence appear to conflict with the direct empirical evidence). When a fairly large body of direct evidence appears to conflict with reason, logic, and/or indirect evidence, this conflict forms that basis for generating new testable hypotheses. Once they’ve been tested, we may gain a more nuanced understanding of the phenomenon, we may learn that our logic led us astray because we had an incomplete understanding of the mechanisms contributing to the phenomenon, or we may discover that the current body of direct empirical evidence is fundamentally flawed in some way. But, our understanding of the phenomenon is updated once those studies have been conducted, not when those hypotheses are generated.

Claiming knowledge (i.e., making affirmative truth claims) about a phenomenon that conflicts with the direct empirical evidence on the topic is, therefore, fundamentally antiscientific. No matter how well you believe you understand the mechanisms underpinning a phenomenon, and no matter how strong you believe your reasoning faculties to be, the logical and reasonable beliefs you arrive at on the basis of indirect evidence and plausible assumptions do not prove or disprove anything. Rather, they provide you with the basis for asking questions in the form of testable hypotheses. Claiming knowledge before those hypotheses are tested fundamentally skips the whole “science” part of the process. You’re taking a process used for generating questions, and pretending like it generates answers. It’s intellectually lazy, epistemically dishonest, and the antithesis of scientific reasoning.

Telling the wolves from the sheep

The main reason I care so deeply about this topic (the value of boring and rigorous empiricism as a primary epistemic commitment – not training volume) is that I fundamentally think that the greatest value a “science-based” content creator can provide to their audience is modeling the process of scientific reasoning, which is inherently empirically driven and epistemically modest.

If you make a claim, the strength of your claim should scale with the strength of the evidence. If there’s a large body of direct empirical evidence about a particular outcome, you can make strong claims. If there’s a small body of direct empirical evidence about a particular outcome, you can make more tentative claims, while acknowledging the uncertainty inherent to small bodies of research. If there’s no direct empirical evidence supporting your claim, it’s fine to state the belief you currently hold on that topic, but you should acknowledge that your belief is speculative, because there’s not any evidence to specifically validate or refute it. 

When you do this, you’re teaching people a valuable skillset that will help them sniff out a lot of bullshit.

Most people have little-to-no formal scientific training. Few people have the skillset required to find the research on a particular topic, fewer still have the skillset required to read and understand the research once they find it, and fewer still have the time or interest to dig into all of that research, even if they could find, read, and understand it. So, when most people have a question about a particular topic, or if they want to understand a topic better, they’ll typically seek out advice or explanations from someone who appears to have expertise about that topic.

So, if someone is consuming your content because they view you as an expert on a particular topic, you’re providing an answer to their question, but you’re also modeling the way that an expert would go about answering their question.

The answer itself isn’t particularly transferable – if I tell you how much training volume to do, that answer won’t extend to topics that aren’t related to training volume. But, the process of answering the question is transferable. If they see that an expert on a particular topic is primarily focusing on the empirical data most directly related to the topic of the question, contending with the strengths and weaknesses of the data, and ultimately providing a tentative answer that scales with the strength of the evidence, you’re providing a good example of how an expert goes about answering questions more generally. This will help inform and update their heuristics about what expertise looks like and sounds like, and ultimately help them to more reliably discriminate between credible and noncredible sources of information on other topics.

On the flip side, rationalism is the language of pseudoscience and charlatans. If you make strong claims based on indirect evidence and presumed mechanisms connected with a bit of basic logic, you’re modeling a process of generating conclusions that others can exploit.

The next time you encounter pseudoscientific claims on social media, I ask that you take a brief moment to analyze why someone might find those claims to be convincing.

Here’s an example that shouldn’t be too controversial: to maximize muscle growth, you should take BCAAs before your workout. BCAAs stimulate muscle protein synthesis and reduce protein catabolism. This will help you recover faster from training and build more muscle.

Every discrete claim in that statement is true:

These claims contain two implicit arguments that are almost so obvious they don’t need to be articulated:

Argument 1:

Premise 1: Hypertrophy is ultimately the result of the net accretion of muscle protein.

Premise 2: By definition, net accretion of muscle protein requires protein synthesis to exceed protein breakdown.

Conclusion 1: Therefore, interventions that increase muscle protein synthesis and reduce protein breakdown will increase hypertrophy.

Argument 2:

Premise 1: Interventions that increase muscle protein synthesis and reduce protein breakdown will increase hypertrophy.

Premise 2: BCAAs increase muscle protein synthesis and reduce protein breakdown.

Conclusion 2: Therefore, BCAAs increase hypertrophy.

These are well-constructed, formally valid logical arguments with completely reasonable premises. However, they ultimately lead to a conclusion that’s untrue: it’s now pretty well-established that BCAAs don’t actually help you build more muscle, based on studies that directly test the effects of BCAAs on the outcome of interest. But, if you formed your conclusions based on the apparent validity of the logic supporting the use of BCAAs rather than the direct empirical evidence against the use of BCAAs, you’d have no reason to believe that BCAAs aren’t one of the most effective supplements on the market for promoting muscle growth.

The same is true of basically all of the quack recommendations you’ll encounter … anywhere. This is certainly the case in the health, fitness, and wellness industry. Virtually every supplement, fancy gizmo, fad diet, biohack, etc., is supported by some reasonable-sounding train of logic, and typically some indirect or mechanistic evidence on some level. But, the vast majority either have no direct evidence supporting their efficacy for the outcome they’re supposed to promote, or their efficacy is directly contradicted by the most direct evidence. This also extends to pseudoscience you’ll encounter in other areas as well, from psychology to climate to economics to medicine. Pseudoscience almost always takes the form of reasonable arguments predicated on indirect evidence, used to promote beliefs or practices that are unsupported or even refuted by high-quality direct evidence.

When you model the same methods of evaluating evidence employed by quacks, you may not be signing off on the specific beliefs and recommendations promoted by quacks, but you’re making your audience more susceptible to quack beliefs (or, at minimum, you’re not demonstrating methods of evaluating evidence that would make them less susceptible to quack beliefs) by endorsing the epistemic processes that produce and support quackery.

I should also note that I’m not questioning anyone’s intentions here. I truly believe that most rationalists in the fitness industry who present themselves as “evidence-based” or “science-based” actually believe the things they say, and I truly believe that they think the process they use to arrive at conclusions and recommendations is “scientific.” So, I’m not saying they’re charlatans. But, I am saying that the way they interpret and discuss research – the way they model expertise – is indistinguishable from the way that quacks and charlants interpret and discuss research for most consumers of fitness content. If you’re modeling intellectual practices that can be easily exploited by quacks who want to be able to present themselves as legitimate experts, while simultaneously branding yourself as an “evidence-based” or “science-based” creator, you may not be directly supporting specific quacks, but you are lending support for quackery writ large by making it easier for quacks to adopt similar “science-based” branding, and you’re doing your audience a grave disservice as a result. 

Of course, the veneer of empiricism can be exploited by charlatans as well. But, it’s much, much harder. The number of obviously bullshit claims that can be supported by logical-sounding arguments is at least an order of magnitude larger than the number of bullshit claims that can be supported by a large volume of direct empirical evidence. The next time you find an influencer promoting scams while citing systematic evidence syntheses that directly support their claims, send me a link. They’re a rare breed.

So, just to state this clearly: I don’t particularly mind that some people think lower volumes are better for muscle growth. I don’t mind that people base beliefs on logical arguments instead of direct empirical evidence (again, that’s the default way most people form beliefs about most things, assuming they’ve thought enough about a topic that their beliefs aren’t just based purely on vibes or social conformity). But, it does bother me when people present themselves as credible experts on a topic, use “sciencey” brand aesthetics to give themselves a greater air of legitimacy, and then promote methods of reasoning that are fundamentally antiscientific, and that lend cover and legitimacy to quacks and charlatans.

Now that this brief (ha ha) aside about epistemology is finished, let’s return to the nominal topic of this article: training volume. The first part of this article primarily addressed the reasons why you can be confident that higher training volumes do actually promote more muscle growth, and the main weaknesses of the primary arguments against that position. So, to wrap things up, I’m just going to cover a handful of additional volume-related topics, presented as a (somewhat) rapid-fire FAQ.

Volume FAQs

Where’s the conservative volume limit?

For most of this article, I’ve been writing with pretty general terms (i.e. “higher volumes” and “lower volumes”), I now want to focus on a more concrete question: what is the highest level of volume I’d feel quite confident recommending in a general sense?

Short answer: somewhere around 25 sets per muscle group per week.

The reason for this answer is pretty straightforward: data density drops considerably at volumes exceeding 25 sets per week. There are a lot of studies testing levels of volume at or below 25 sets per week, and only 5 studies testing levels of volume exceeding 25 sets per week. In those 5 studies, even higher volumes do still look promising, but my confidence in any recommendation scales with the quantity of data supporting it.

Incidentally, this is also the point past which the authors of the Pelland meta-regression recommend caution when interpreting the results of their model: “caution is warranted as few studies have explored ~25+ ‘fractional’ weekly sets. Therefore, future research may wish to explore these higher volumes to better inform the dose-response and potential plateau point.”

Where’s the actual volume limit?

The only honest answer is that we don’t yet know. As mentioned above, we only have 5 studies testing volumes that exceed 25 sets per muscle group per week. Furthermore, the highest volume (averaged over the course of the training program) tested in any of these studies was 45 sets per week. In that group of 5 studies testing volumes in excess of 25 sets per week, the evidence does still support greater hypertrophy with larger training volumes – i.e., 45 sets appear to promote larger gains than 25 sets.

I know that everyone is expecting volume to have an “inverted U” relationship with hypertrophy: increased hypertrophy with increased training volume to a point, past which further increases in volume will yield less total hypertrophy, because lifters will exceed the level of volume they can recover from. I count myself as part of “everyone” in that statement.

However, based on the current research, we haven’t found the inflection point of that “inverted U” yet. To this point in the literature, every time higher and higher volumes have been tested, the results have suggested that the volume “limit” may be even higher yet. In other words, the number of sets required for growth to plateau and begin decreasing currently appears to be some number exceeding 45 sets.

With that said, one or two null or negative results could change that pretty quickly. That’s the main reason I’m only quite confident about levels of volume up to about 25 sets per week. Basically, if one or two studies are published finding more growth with 10 sets per week than 20 sets per week, we could still be pretty confident that 20 sets will lead to more growth than 10 sets in general, because we already have enough studies in the 10-20 set range that one or two null or negative results wouldn’t significantly shift the balance of evidence. However, if one or two studies were published finding more growth with 30 sets than 45 sets, that would shift the balance of evidence quite a bit, and it may start looking like the high point in the “inverted U” is somewhere around 30 sets per week.

However, I should also acknowledge that it’s entirely possible that the limit might be considerably higher. I don’t think we have enough data to have a high degree of confidence about what happens with volumes exceeding 25 sets per week, so I certainly don’t think we have enough data to have a high degree of confidence about what happens with volumes exceeding 45 sets per week.

Isn’t the consensus “evidence-based” volume range 10-20 sets per week?

In the recent volume discourse, I’ve seen volume recommendations exceeding 20 sets per week framed as a crazy new fad, or a transgression of some long-held scientific consensus. This criticism is predicated on the idea that we’ve “known” for decades that the optimal volume range was 10-20 sets per week, and suggestions that even higher volumes may be beneficial goes against an established scientific principle that’s supported by loads of evidence.

However, that’s simply not true.

What is true is that plenty of “evidence-based” fitness influencers have said that 10-20 sets was the optimal volume range for hypertrophy, but there’s never been particularly strong evidence suggesting we couldn’t achieve additional growth with volumes exceeding 20 sets per week.

Realistically, I know exactly when the “10-20 sets” recommendation became the standard “evidence-based” recommendation (I’m not necessarily saying that this was the first time it was ever used, but it is when it became ubiquitous). It was a lazy interpretation of Schoenfeld’s 2017 volume meta-analysis. At the time, there were only 15 studies investigating the effects of varying levels of training volume, and rather than performing a formal meta-regression, the researchers quantified the effects of training volume in three “buckets”: <5 sets per week, 5-9 sets per week, and 10+ sets per week. With this bucketing approach, it didn’t make sense to include a “bucket” for, say, 20+ sets per week, because there had only been two studies investigating volumes at or above 20 sets per week – you can’t really perform a (meaningful) meta-analysis of just two studies.

This meta-analysis generally found that higher volumes led to more growth, with 10+ sets per week outperforming <10 sets per week. So, people just started recommending 10-20 sets per week, since the “10+” bucket mostly included studies with volumes of 10-20 sets per week, and because (I suspect), “10-20” felt like a more “sciencey” recommendation than “10+”. But, notably, this meta-analysis didn’t actually provide any evidence that people couldn’t achieve more growth with even higher volumes – it simply suggested that maximizing hypertrophy required at least 10 sets per week.⁵

We can also look at the research on the topic sequentially to determine whether there was ever a strong empirical case for the benefits of volume maxing out at 20 sets per week.

To do this, I pulled all of the studies where at least one group trained with a volume of at least 20 sets per week, and ordered them chronologically so that we can take rolling snapshots of the state of the research on the topic. In all of these studies, I only included adjacent volume groups where the group performing more volume was performing at least 20 sets per week. So, for example, in our very first study (Ostrowski, 1997), there were three groups of subjects doing 5, 10, and 20 sets of triceps training per week. For this analysis, I dropped the results from the 5-set group, since the adjacent comparison between 10 sets and 20 sets is what we care about here.

Instead of an unwieldy table, I think it would probably be easiest to just share scatterplots of the data with each new study that was published. So, here’s how the case for high training volumes (20+ sets versus <20 sets) looked when the very first study on the topic was published:

With just one comparison from one study, the “10-20” set range actually looks quite defensible. In the very first study on the topic by Ostrowski and colleagues, 10 and 20 sets led to virtually identical gains in triceps muscle thickness. But, let’s see what happens when we add a second study investigating training volumes exceeding 20 sets per week. Chronologically, the next study was by Radaelli and colleagues, from 2015:

With the addition of one additional study, we’re now seeing a clear positive slope, and it certainly appears that >20+ sets per week yields more growth than 10-20 sets per week. I’ll also pause here to note that the “10-20” set range may have felt quite plausible because, from 1997 to 2015, there was only one study that included a training volume of at least 20 sets per week (Ostrowski), and it did suggest that 20 sets yielded the same results as 10 sets. So, a single study doesn’t tell you much, but a single study was literally all that we had for 18 years. But, as soon as there was a second study on the topic, the “10-20” set range was immediately far less defensible.

But, maybe a third study will turn things around. This one was by Amirthalingam in 2016:

It still looks like there’s an overall positive trend, but a simple “vote counting” approach to analyzing these three studies could tell a slightly different story. The Amirthalingam study actually provides some of the strongest evidence against the idea that 20+ sets will yield more growth than 10-20 sets. This study assessed changes in thigh muscle thickness after training with 14 vs. 24 sets per week, and it actually observed more growth with 14 sets (4.9%) than 24 sets (2.1%). So, you could make the case in favor of volumes exceeding 20 sets per week based on the aggregate positive trend, or you could take the more skeptical approach, and note that, as of 2016, we had one study suggesting that 20+ sets yielded more growth (Radaelli), one study suggesting that 20+ sets yielded less growth (Amirthalingam), and one study finding no difference between 10 and 20 sets (Ostrowski).

The next study on the topic was by Shoenfeld and colleagues, published in 2019. It was another generally positive finding in favor of higher training volumes:

In retrospect, this now looks like the point past which the case for “10-20 sets” became untenable, at least when we analyze the volume studies included in the Pelland meta-regression. However, it’s worth taking a brief historical detour, because this time period was actually a bit of a flash point when the low volume position had considerable momentum. Around this time, two studies were published in quick succession that appeared to show that hypertrophy was actually maximized with just 5-10 sets per week:

Evidence of a Ceiling Effect for Training Volume in Muscle Hypertrophy and Strength in Trained Men – Less is More?, and Evidence for an Upper Threshold for Resistance Training Volume in Trained Women by Barbalho et al.

Below you can find the graphs of the hypertrophy results from the study in female lifters. The male volume study had virtually identical results:

If you clicked on the links above to read these studies, you may have noticed that they’ve both been retracted. One of them was retracted because the authors actually admitted that the data was fabricated (the senior author said that an undergraduate student responsible for transcribing the data in the male volume study “made much of the data” because he was “too busy to do the job.”), and the other was retracted due to a long list of data anomalies in that study and multiple other papers from the same research team.

But, before these studies were retracted, they were incorporated as a cornerstone of the case against high training volumes. These were pretty long studies (24 weeks) with trained subjects, the subjects trained each muscle group just once per week, strength gains followed the exact same pattern as hypertrophy outcomes, and hypertrophy measures were allegedly taken 72-120 hours after the final training session (instead of the typical 48-72 hours). So, they served as the strongest evidence in favor of several positions you still frequently see today:

1. Hypertrophy benefits top out after just ~5 sets per workout. Anything past that is “junk volume.”
2. Strength gains scale almost perfectly with hypertrophy, even when varying training volume.
3. The apparent benefits of higher volumes are actually just due to swelling in shorter studies that take post-training measures of muscle size too soon after the last training session.

When these studies were finally retracted, the strongest evidence in favor of all of those positions evaporated. The people who held these positions silently stopped citing these studies, but they didn’t actually change (or even soften) their positions. They simply laid low for a year or two, and cooked up new justifications for their positions that could accommodate the severe reduction of direct evidence (and even reasonably strong indirect evidence) supporting them. As is often the case, the initial publication of fake research had a much larger impact on people’s beliefs than the eventual retraction of the fake research. But, I cannot overstate the degree to which these (now retracted) studies played a critical role in the development and initial propagation of the current batch of volume-skeptic arguments.

Now, let’s get back to our timeline. We can just quickly run through the most recent studies to catch up to the present day. The next non-retracted study was by Evangelista and colleagues in 2021. Let’s add it to the graph:

Next, Brigatto et al. from 2022:

Next, Aube et al. from a bit later in 2022

Finally, Enes et al. from 2023:

In total, the average hypertrophy observed with 9-19 sets per week was 4.41%, and the average hypertrophy observed with 20+ sets per week was 9.02%. Furthermore, of the 19 distinct within-study comparisons in these high-volume studies (where the higher-volume group is doing at least 20 sets), 16 (84%) observed nominally more growth with higher training volumes, while only 3 (16%) observed nominally more growth with lower training volumes (if the graph below is confusing, it may be worth giving this article a quick skim for its discussion of within-study comparisons).

So, when looking back on this body of research, I think we can pretty comprehensively say that there was never a long period of time where there was a strong empirical case in favor of 10-20 sets being the clearly “optimal range” of volume for hypertrophy. The period of time when you could make the strongest case was from 1997-2015, but that case would need to be predicated on precisely one study. In 2016, you could make a significantly weaker case based entirely on a “vote counting” approach to evidence synthesis. The period from 2019-2021 was really a no-man’s land (the period of time between the publication and subsequent retraction of the Barbalho studies) – you could make a case for hypertrophy being optimized with 5-10 sets per week, or with 30-45 sets per week. This is really when 10-20 sets became solidified as the default “evidence-based” recommendation, but not on the basis of affirmative evidence specifically supporting it. It was just a popular “centrist” position, and felt more sciencey than just recommending “10+” on the basis of the Schoenfeld meta-analysis.

So, the volume recommendation of 10-20 sets was the default recommendation made by many “evidence-based” fitness influencers, but it was never a recommendation that was directly supported by much actual evidence (i.e., there was never particularly strong evidence that 10-20 sets per week maximized hypertrophy, with volumes exceeding 20 sets per week failing to yield additional growth). 

Is it even possible to do 30+ sets per week for every body part?

On one hand, it’s all well and good to point at the published research and say that more volume is better. On the other hand, it sure seems like training would need to become your part-time job to actually employ a training volume of 30+ sets per body part per week. And, more importantly, there are concerns that it may be tough to actually recover from such high volumes for all body parts.

This is a concern that gets at one of the primary weaknesses of the volume literature: many of these studies have used training programs that only include a handful of exercises and target a handful of muscles.

Now, that’s not universally true. For example, the high-volume study by Schoenfeld used a program that included 5 sets of flat barbell bench press, barbell military press, wide grip lateral pulldown, seated cable row, barbell back squat, machine leg press, and unilateral machine leg extension, performed three times per week. It neglects the hamstrings, glutes, and calves, and serious bodybuilders would probably want to include some more shoulder exercises and direct arm training, but it would be training most major muscle groups. Similarly, the high-volume program employed in the Brigatto study can be seen below. It’s also not perfect – you’d probably want some more work for the shoulders, calves, and glutes, for instance – but it does include pretty high volumes for most major muscle groups.

But on the flip side, there are also studies like Aube (subjects just did squat, leg press, and low-volume glute-ham raise), and Enes (similar – just high-volume quad training and low-volume hamstrings training), that only involve intensive training for a much smaller handful of muscle groups.

So, you might argue that optimal training volumes may be lower in the real world. In part due to lifestyle considerations (i.e., just being able to carve out enough time to do that much training), and in part due to global recovery concerns – if you’re training more muscle groups than are trained in the high-volume studies, would you still be able to recover and adapt? I’ll mostly be focusing on recovery concerns here (feasibility related to schedules and lifestyle considerations will be discussed later).

To start with, survey data from competitive bodybuilders suggests that volumes of 30+ sets per week for most major muscle groups aren’t the norm, but they aren’t terribly uncommon either. A 2022 study reported that over 50% of the bodybuilders in their sample trained most muscle groups twice per week, performing 2-3 exercises per muscle group, and 3-4 sets per exercise. So, this suggests that a volume of 12-24 sets per muscle group is pretty typical. However, deviations from these norms tended to skew upward. In other words, around 20% of respondents had higher training frequencies (3+ weekly sessions per muscle group), around 30% performed 4-5 exercises per muscle group, and around 20% performed 5-6 sets per exercise. This suggests that volumes of up to around 36-40 sets per muscle group per week aren’t too far outside the norm. For example, with the typical number of exercises per muscle group (2-3 per session) and the typical number of sets per exercise (3-4 sets), but with a frequency of 3 sessions per muscle group per week, you’re looking at a volume range of 18-36 sets. Or, with the typical frequency (2 sessions per muscle group per week) and typical number of sets per exercise (3-4 sets), but a slightly higher number of exercises per muscle group in each workout (4-5 exercises, instead of the typical 2-3), you’re looking at a volume range of 24-40 sets. Furthermore, the authors note that these figures may still undercount the total volume performed per muscle group, since the survey data potentially underestimates the impact of exercises targeting muscle groups: “It should be noted that the training volume could even be higher because exercises targeting the muscle groups listed in the survey might engage more than one of these muscle groups. For example, dips might have been listed by respondents as an exercise targeting the arm muscles (i.e., triceps brachii) but the chest muscles (i.e., pectoralis major) would also likely contribute to the performance.”

I’ll also note that the more direct research on the topic doesn’t give much validation to the concern that performing 30+ sets for multiple muscle groups isn’t generally manageable. There’s not a clear pattern in the data suggesting that high volumes are only beneficial when subjects exclusively train one or two muscle groups, but that high volumes fail to yield better results when more muscle groups are trained. Rather, we tend to see pretty similar results in both types of studies.

However, I’ll also admit that I partially share this concern purely on a gut level, and my bias is that really high volumes for all muscle groups may not be feasible. This bias is primarily informed by personal experience (I’ve consistently run into issues when I’ve tried to push volume too high for too many lifts at once) and general “gym wisdom.” The concept of really high-volume training isn’t foreign to bodybuilders, but the default recommendation is to employ it in the form of “specialization cycles” for just one or two muscle groups at a time (while training the rest of your body with lower volumes).

If you did want to make the case that high volumes for all body parts may not be feasible or beneficial, here’s some of the arguments or indirect evidence you might lean on.

1. You might note that most of the research is performed on fairly young subjects (average age is around 25 years old, and most studies lean pretty heavily on college students for most of their subject pool). So, even if high volumes for multiple body parts is feasible for these subjects, it may not be feasible for people in their 30s or above.
2. You might argue for some type of non-local fatigue effect – the more muscles you train with high volumes, the more that will lead to generalized fatigue that will negatively influence your ability to generate tension and achieve a large training stimulus in other muscles.
3. Relatedly, you might argue that the inflammatory response that would be caused by training too many muscle groups with high volumes could contribute to generalized overtraining.
4. You might make a case that global hypertrophy is constrained in some way.
   1. There’s some evidence that hypertrophy in muscles you do train may contribute to atrophy in muscles you don’t train. So, training a muscle may exert some small atrophic stimulus on other muscles. If this effect is real, and if it scales with the magnitude of the training stimulus for each muscle, then at some point the increased hypertrophic stimulus for each muscle may be “canceled” out by the increased atrophic stimulus from other muscles as volume increases.
   2. There’s some muscle protein synthesis (MPS) research suggesting that per-muscle MPS may be a bit lower when training multiple muscle groups. You might remember that people used to claim that 20g of protein was sufficient to maximize MPS post-workout. That was due to two papers comparing 20g to 40g of post-workout protein, and finding no significant difference between the two (one, two). But, both of those studies only involved quad training. When another study was published including full-body training, 40g of protein did lead to significantly more MPS than 20g. Furthermore, the rate of MPS (fractional synthetic rate) in the quads was lower (0.5-0.6% per hour) with full-body training than just quad training (0.7-0.8% per hour). So, you could interpret this to mean that training more muscle groups may just increase protein requirements, but you could also interpret this to mean that training multiple muscle groups reduces MPS in each muscle group you train. Again, if this effect is real, and if it scales with the magnitude of training stimulus for each muscle you train, then at some point the anticipated increased MPS response from training one muscle with higher volumes may be canceled out by the effect of training other muscles with higher volumes.

I’ll note that if you find any of those arguments compelling, they don’t constitute affirmative evidence that you can’t train multiple (or all) muscle groups with high volumes. Rather, they give rise to concerns that can give rise to testable hypotheses, but they don’t constitute strong evidence until those hypotheses are tested.

But, on the flip side, it’s also possible that the anecdotal evidence giving rise to my personal bias, and the general recommendations to confine really high volumes to “body part specialization blocks,” is in part a product of exercise selection.

Thinking back, all of my negative experiences with really high training volumes had one of two commonalities:

1. I did way too much overhead or incline pressing, which aggravates an old shoulder issue I developed playing baseball.
2. I tried to really push my squat and deadlift volume simultaneously, leading to lower back tightness and fatigue that I couldn’t shake.

But, when that would happen, I was always fine if I just cut back on squats and deadlifts in favor of leg press, back raises, and hamstring curls, or if I dropped some pressing in favor of flyes, delt raises, and triceps extensions. It didn’t really seem like an issue with muscular recovery – I was just prioritizing exercises that frequently cause me joint issues or a lot of generalized fatigue (deadlifts especially) when done in excess. For example, when I look at the high-volume training protocol used in the Brigatto study, it actually looks extremely reasonable to me, but I know that if I subbed in any type of deadlift in place of leg curls, or if I added overhead press to that program, I’d have major issues. However, those major issues would be due to high-volume deadlifts or high-volume overhead press specifically – not high-volume training per se.

The specific exercises that cause issues and the specific joints or pieces of connective tissue that give you fits may be different for other people, but I suspect this overall experience is generalizable to some degree.

Similarly, I suspect a lot of negative anecdotes come from people who dive into high-volume training headfirst without knowing what to expect. If you immediately increase your training volume from, say, 10 sets per body part per week to 25 or 30, one of two things will typically happen. The first (less common) possibility is that you get a little muscle strain or develop some type of joint or connective tissue inflammation because you tripled your workout overnight, and your tissues weren’t ready for it. The second (almost universal) possibility is that you feel like you’re dying for about 2 or 3 weeks before you adapt (but many people abandon it after 2 or 3 weeks before they adapt, because they think they’re going to be sore from head-to-toe and brutally fatigued for as long as they keep their volume high). Either way, you’re going to have a rough time. Both of these outcomes can typically be avoided by simply ramping up your volumes more gradually (i.e., over 3-6 months, instead of overnight). 

Overall, I don’t think we can fully rule out the possibility that high volumes are primarily effective if used for just a handful of muscle groups, because there haven’t yet been any studies that train all major muscle groups with high volumes. However, as mentioned above, the current evidence doesn’t validate this concern – we see pretty similar results from studies that only train one or two muscle groups with high volumes, and from studies that train five or six muscle groups with high volumes. At this point, it does appear that you can still benefit from training all (or at least most) muscle groups with quite high volumes.

With that said, I do believe that full-body, high-volume training may carry a bit more risk, and require a bit more adaptability. You may find that certain exercises don’t cause you any issues when you perform them for 10 sets per week, but they do cause issues when you perform them for 30 sets per week. Similarly, you may just have one or two joints or tendons that do present you with something like a hard volume cap for certain muscles, which you’ll need to be mindful of if you try to ramp up your volume across the board.

Personally, I still have a soft spot for the concept of body part specialization blocks. If nothing else, I think they can serve as a low-risk way to dabble with high-volume training, so that you can decide whether you like it, and whether it seems to work well for you before you fully take the plunge.

Does everyone grow more with high training volumes?

Zooming out significantly, I think the “volume debate” is noteworthy simply for how long it’s existed, and for how popular the low-volume position has continued to be.

When we look at the research, it certainly appears that higher training volumes are simply better than lower training volumes, and it’s not particularly close. Most of the time when we’re dealing with such a clear difference in results, general opinion more or less coalesces around the “right answer,” even before we have a ton of research on the topic. For example, there’s been a near-universal consensus that higher protein intakes are helpful for muscle growth, and that heavier loads tend to be better for strength gains for a long time – well before there was a bunch of published research on either topic. Typically, we need research to help us identify and build confidence in smaller effects, but effects that are this large tend to be extremely obvious. So, it’s worth giving this riddle a bit of thought. Why is the topic of training volume an area where there’s widespread disagreement, rather than consensus?

One potential answer (and the one that’s favored by the low-volume crowd) is that the research itself is wrong. Higher training volumes don’t actually cause more muscle growth. They just cause more inflammation, swelling, edema, etc. That’s what the first half of this article focused on.

A second potential answer is more sociological. People interested in maximizing their strength are generally people who like lifting heavy weights. So, it’s convenient that heavier loads are better for maximizing strength gains, but it also wouldn’t take much to convince strength athletes that lifting heavier is better for them (i.e., it’s a belief they’d already want to hold). With protein, companies have been marketing high-protein powders and products for decades. This pads their bottom line, but all of that marketing also promotes the view point that higher protein intakes are a good thing. So, there were market forces that incentivized creating consensus around the topic. With training volume, on the other hand, things are a bit different. “Get better results in less time” is, and always has been, a compelling sales pitch – just ask Arthur Jones and Mike Mentzer (the two people who are probably most responsible for initially popularizing low-volume training). I don’t doubt that they believed in the ideas they promoted, but they did both make a pretty penny by promoting low-volume training, because their sales pitch was inherently appealing. In other words, the enduring popularity of low-volume training may simply be due to the fact that it’s typically more profitable to promote low-volume training, because it’s inherently appealing to believe that you can maximize your results with relatively low training volumes.  

A third potential answer, though, is quite a bit simpler: high-volume training doesn’t work for everyone.

For example, let’s just assume that higher training volumes linearly increase muscle growth in 50% of people, but have no impact on muscle growth past some fairly low level (maybe 10 sets per week) in the other 50%. What might we see if we researched the effects of training volume in this population?

Well, with training volumes below 10 sets per week, more volume would lead to more growth for everyone. But, if volume increased from 10 sets to 20 sets, half of the subjects would achieve twice the gains, and half of the subjects wouldn’t see any benefit. But, when we analyzed results at the group level, it would look like 20 sets caused 50% more growth than 10 sets. The same trend would continue as training volumes continued increasing. So, as training volume continued increasing, average growth would increase, but variability in growth responses would increase as well. 

Once again, I’ll admit my bias early on: I personally believe that something like this is happening. Anecdotally, I’ve known a lot of people who’ve gotten better results with higher volumes, and I’ve known plenty of people who’ve gotten better results with lower volumes. I’ve personally trained a lot of people whose results improved as volumes increased, and a lot of people who thrive with lower volumes and plateau (or even regress) on higher volumes. Amongst world-class bodybuilders, there were people who built their physiques with high volumes, and people who built their physiques with lower volumes. To be clear, I certainly don’t think this is a strictly binary response (i.e., you either respond really well to higher volumes, or you don’t respond at all to higher volumes, with no middle ground), but I do strongly suspect that “volume responsiveness” varies considerably between individuals.

But, I’ll also readily admit that the evidentiary case for this perspective is sparse – in no small part because most research isn’t designed to test variability in training responses.

However, here are some bread crumbs:

First, I checked some of the studies that have tested some of the highest training volumes in the literature, and wanted to see whether hypertrophy responses actually become more variable as volumes increase.

And…they certainly might.

For example, here’s the relationship between training volume and hypertrophy standard deviations from Brigatto:

Here’s Schoenfeld:

And here’s Enes:

So, this gives us some indication that hypertrophy responses may become more variable as training volumes increase, which is what you’d expect to see if higher training volumes only “work” for some people: in these studies, as volume increased, hypertrophy responses generally became a bit more variable (though, admittedly, the trends weren’t particularly strong).

However, most volume studies don’t report enough data to calculate variability in hypertrophy responses, so these may just be three studies that are non-representative. Furthermore, more volume tended to cause more growth in all three of these studies, and standard deviations tend to increase as means increase anyways. There are also methodological considerations for determining whether you’re actually seeing “true” response variability or just random noise (important topic, but beyond the purview of this article) that weren’t accounted for here.

But, we can do slightly better. Two studies come to mind that tested different levels of training volume using within-subject designs.

The first is by Hammarstom and colleagues. It compared the effects of doing about 6 vs. 18 sets of quad training per week in a group of untrained subjects (one leg did 6 sets per week, and one leg did 18). For 18 out of 34 subjects, hypertrophy responses were similar in both legs (differences between legs didn’t exceed the “smallest worthwhile change” – basically the cutoff point between a “trivial” and “small” effect size). However, 13 subjects experienced meaningfully more growth with higher training volumes, and 3 experienced meaningfully more growth with lower volumes. Strength outcomes followed a pretty similar pattern. In both cases, the subjects who benefited more from higher training volumes were the subjects who experienced larger increases in muscle RNA, which is a proxy for ribosome biogenesis.

The second was a study by Damas and colleagues. This study tested the effects of different training frequencies (frequencies of 2, 3, and 5 times per week), but each training session employed the same volume, such that a frequency of five times per week coincided with 2.5-times more volume than a frequency of 2 times per week. In this study, the number of subjects achieving better results with higher vs. lower training volumes was more balanced (5 subjects experienced at least a 5% larger increase in vastus lateralis CSA with less volume and frequency, and 5 subjects experienced at least a 5% larger increase in vastus lateralis CSA with more volume and frequency). But, the difference in results was typically larger for the subjects who achieved better outcomes with higher volumes. In other words, subjects that got better results with lower volumes had one leg that grew 5-7% more with lower volumes, and the subjects that got better results with higher volumes had one leg that grew ~8-15% more with higher volumes.

It’s also noteworthy that one of the most well-known studies documenting variability in resistance training responses was a study employing a pretty high-volume protocol. A pair of papers from Bamman and Petrella documented the spread of strength and hypertrophy responses in a group of 66 subjects following 16 weeks of quad training. The subjects included younger (20-35 years old) men (N=21) and women (N=16), and older (60-75 years old) men (N=14) and women (N=15). The training protocol consisted of 27 sets of quad training per week: three sets of squats, three sets of leg press, and three sets of knee extensions performed to volitional failure with 8-12RM loads, three times per week.

On average, these subjects experienced considerable hypertrophy – mean vastus lateralis fiber cross-sectional area increased by about 28%. However, there was a huge spread of results. The bottom ~25% of subjects experienced an average of approximately zero hypertrophy, while the top ~25% of subjects experienced a mean 58% increase in fiber CSA – more than double the overall average response. 

The researchers found that the high responders were more likely to experience a robust satellite cell response, and experience larger elevations in the expression of anabolic growth factors (MGF, IGF-1, and myogenin). Of note, each age/sex combination was included in each response cluster (i.e., there were younger men, younger women, older men, and older women who were high responders, moderate responders, and low responders), so we can’t just attribute the variability in hypertrophy responses to sex or age.

It’s not entirely clear why only some people would benefit more from higher training volumes. Enhanced ribosome biogenesis is one tentative possibility, but that just leads to more questions (i.e., why do higher training volumes provoke an increased ribosomal response in some people and not others?). The same applies to satellite cell and anabolic growth factor responses (i.e., why do high training volumes lead to a large increase in satellite cell and anabolic growth factor responses in some people but not others?).

Another possibility may be differences in fiber type distributions. Type II fibers fatigue more quickly than type I fibers. So, if you do 10 sets targeting a single muscle group in a training session, but your type II fibers are quite fatigued after the third set, they may receive a smaller additional stimulus from sets 4-10. If you have proportionally more type II fibers, this could mean that the total additional stimulus you receive from sets 4-10 would be quite small. But, if you have proportionally more type I fibers, sets 4-10 may still be providing a very robust hypertrophic stimulus for the majority of your muscle fibers. Furthermore, type II fibers are more susceptible to muscle damage than type I fibers, so you’d generally expect people with more type II fibers to have larger issues recovering from high training volumes. 

Another possibility is that genotype influences volume responsiveness. For example, one study found that people with the II ACE genotype were more likely to benefit from higher training volumes than people with the DD ACE genotype (though that study only assessed strength gains). 

Of course, all of this is quite speculative. We still need more research to determine whether there is actually inter-individual variability in terms of volume responsiveness. And, if there is, we need more research to determine the cause of this variability.

But, I personally believe that this is the primary reason why higher volumes lead to considerably more muscle growth on average, but a large contingent of people is still convinced that lower volumes produce equal or better results. I think it’s quite likely that higher training volumes cause different responses in different people – some people grow more, some people don’t experience much of a difference, and some people may even grow a bit less. In other words, two things can be true:

1. The research is accurate, and higher training volumes tend to cause more hypertrophy on average.
2. Higher training volumes don’t cause more growth for a lot of people.

Why didn’t higher training volumes work for me?

As discussed above, I think that one distinct possibility is that higher training volumes don’t work for you in some generalizable sense.

However, another distinct possibility is that you decreased your level of per-set effort when you increased your training volume.

What we see in the research is that higher volumes lead to more hypertrophy when people put a very high degree of effort into every set. Almost all of the hypertrophy studies included in the Pelland meta-regression either involved exclusively training to failure, or a mixture of failure and non-failure training.

But, in practice, I think many people struggle to motivate themselves to put a high degree of effort into every set when they’re training with high volumes. You might say you’re aiming for 1-3 reps in reserve, but that slowly drifts up to maybe 4-6 reps in reserve over time. And, to be clear, I’m not saying you absolutely need to train to failure on every set of every exercise. But, I am saying that if increased volume comes at the expense of maintaining a high degree of effort on each set, you shouldn’t necessarily expect the results of the volume literature to match the outcomes you achieve in practice.

I think many lifters have the exact wrong idea about the ecological validity of research in the field. People think that research may not translate to the “real world,” because lifters in the “real world” train so much harder than research subjects in a study. But, that is the exact opposite of the truth in almost all cases, especially when a training protocol dictates that subjects need to train to failure.

A major difference between a research context and the “real world” is that, in the lab, you generally have at least two or three highly caffeinated Masters students providing “robust verbal encouragement” on every set (i.e., yelling at you as loud as they need to yell to ensure you take a set all the way to failure, if that’s what the research protocol dictates). There might also be some cash (or extra credit in one of your classes) on the line for completing the study, but you’ll be dropped from the study if you don’t adhere to the protocol (which typically includes training to failure). This combination of incentives, peer pressure, and raw decibels is a very effective way to get way higher levels of effort out of research subjects than they’d typically give if left to their own devices.

So, I think that’s the primary reason why some people get disappointing results from higher training volumes. Instead of doing more sets with the same level of effort per-set, they go from doing a smaller number of high-effort sets, to a larger number of much lower-effort sets. In the volume literature, we see more hypertrophy when people do a larger number of very high-effort sets.

Why don’t we see people doing 40+ sets per muscle group in the “real world?”

One common criticism of the volume literature is that we have a few studies where training with 40+ sets appears to lead to more muscle growth, but don’t see (many) people training each muscle group with 40+ sets per week in the “real world.” If really high volumes do actually yield more growth, it’s assumed that competitive bodybuilders (or other athletes who would benefit from maximizing their muscularity) would already be training with extremely high volumes. So, what’s going on here?

I think the most obvious explanation in most cases is simply one of time constraints: it takes more time to train with higher volumes, and most bodybuilders aren’t “full-time” bodybuilders – they still have lives outside of the gym, day jobs, and other commitments. Maybe they could get 20% better results (for example, building 1.2kg of muscle instead of 1kg of muscle over the course of a year), but they’d also end up spending 50% more time in the gym, and potentially take on a bit of additional injury risk (though, I should note, hypertrophy training is remarkably safe). For most people, that trade-off isn’t worth it.

A second potential explanation is that optimal training volume varies with both chronological age and training age. The plural of “anecdote” is not “evidence,” so take this with a grain of salt, but I know more than a few serious lifters who’ve noted that their volume tolerance decreased over time as they got stronger – myself included. It’s possible that you reach a bottleneck of connective tissue adaptations, such that you’re able to place a bit more strain on your tendons in each workout (for example, if your muscles are twice as strong after 10 years of training, but your tendons are “only” 50% stronger). It’s also possible that you hit a bioenergetic bottleneck – I think it’s easy to underestimate the metabolic demands of resistance training since global energy expenditure isn’t that high over the course of an entire workout, but the local metabolic demands in your working muscles are fairly extreme, and increase linearly with work rate. So, if your strength doubles, the acute metabolic demands on your muscles roughly double during each set you perform. Unless your muscles’ local aerobic and anaerobic capacity increase at the same rate that your strength increases, you’ll simply be capable of generating more metabolic fatigue during each set you perform (and, even if metabolic adaptations did keep pace with strength-related adaptations, larger, stronger muscles still put more occlusive pressure on arteries, reducing oxygen delivery and further amplifying the anaerobic demands of each set). 

So, I think it’s at least plausible that optimal training volume is low for totally untrained lifters, considerably higher for moderately trained lifters, and then it gradually decreases as both training status and chronological age increase. If that’s the case, the research we have on “trained” lifters may be catching the research subjects near the point where they’d be the most primed to benefit from really high training volumes – the “trained” lifters in most of the volume studies had between 0.5-6 years of training experience, and most of the subjects were in their early-to-mid 20s. 

Circling back to a prior section (“Is it even possible to do 30+ sets for every body part?”), it’s also possible that there is some sort of generalized recovery bottleneck when you try to train too many muscle groups with high volumes. This isn’t distinct from the first two possibilities, of course (time demands, and/or optimal volume decreasing with training age). But, I do think it’s noteworthy that strength athletes in more specialized sports often train with extremely high volumes for a much smaller handful of muscle groups. For example, competitive arm-wrestlers, climbers, and grip sport athletes do a tremendous amount of training that targets their wrist and finger flexors, and they often have extremely impressive forearm development as a result (and, in the case of arm-wrestlers, the same applies for their biceps and shoulder internal rotators). 

A final possibility is simply that 40+ sets actually don’t yield more growth than, say, 25 sets. Remember, data density drops precipitously at higher volumes, so it’s still entirely within the realm of possibility that the average optimal volume is less than 40 (or even 30) sets. 

The most honest answer to the question posed by this section header is “we don’t know.” However, I’ll note that this is the honest answer to any question about how far we can extrapolate the research about any training variable.

A common critique of the volume literature goes something like this, “The Pelland meta-regression found that high-volume training led to a ~10% increase in muscle size in just 2-3 months, but we can’t possibly extend those findings into the future. Do you really think people are going to increase their muscle size by 10% every few months just by training with high volumes?” The implication is typically that most of the apparent growth must be due to muscle swelling, rather than “true” increases in muscle size.

This sounds like a fair critique in a vacuum, but when you consider it for more than about two seconds, it becomes pretty obvious that this criticism is, to put it mildly, a pretty selective criticism.

For starters, the same criticism would apply to the proximity to failure literature. This gets us back to “A simple guide to maximize hypertrophy with zero effort” territory. Remember, a prior meta-regression also found that training to failure led to a ~10% increase in muscle size, on average, over a similar period of time. So, if you’re skeptical of studies on high training volumes because you’re uncomfortable with the implication that lifters could get 10% bigger every 2-3 months, you should be equally skeptical of the same implication in studies that involve training closer to failure. Take that skepticism to its logical conclusion, and we’re back to recommending 5 sets with 10RIR as optimal hypertrophy training.

Second, this criticism applies just as easily to low-to-moderate volume training. You wouldn’t expect high-volume training to grow your muscles by 10% every 2-3 months, and you also wouldn’t expect low-volume training to grow your muscles by, say, 5% every 2-3 months.

To illustrate what I mean, a 2018 study sought to explore the limits of human muscularity, and recruited a sample of 95 extremely muscular athletes (consisting of American football players, powerlifters, sumo wrestlers, and shot putters). This group of subjects probably didn’t contain anyone at that absolute pinnacle of human muscularity (i.e., it didn’t include any IFBB pro bodybuilders or professional strongman competitors), but it did contain some extremely muscular subjects. The average fat-free mass index was 25.4, which is in line with elite bodybuilders before the steroid era. And, in this group of subjects, the most jacked of the bunch was estimated to have over 110kg of fat-free mass, and nearly 60kg of skeletal muscle mass.

As a point of contrast, the average untrained male tends to have around 30-35kg of skeletal muscle mass. So, extremely muscular athletes have around twice as much skeletal muscle mass as untrained males. That’s obviously a pretty large difference.

However, if an untrained subject was able to increase their muscle mass by “just” 5% every 2-3 months, it would only take them about 2-3 years to also be one of the most muscular people on the planet. Clearly, this does not happen. In other words, it would be absurd to extrapolate the results from high-volume studies too far into the future, but it would also be absurd to extrapolate the results from low-to-moderate volume studies too far into the future (at least in this manner – assuming you’ll still see x% gains every 2-3 months).

So, this leaves us knocking on the door of one of the most fundamental unanswered questions in the field of exercise science: how should we extrapolate the results of short-term training studies to the “real world,” where people will be training for 10-20+ years, rather than just 2-3 months.

For starters, it helps to characterize the general shape of long-term training adaptations. To this point, it appears that long-term adaptations are best modeled with logarithmic curves. That provided the best fit for two of the rare large datasets that let us model long-term strength adaptations (I’m not aware of anything comparable for hypertrophy): powerlifting results over competitive careers spanning ~20 years, and strength gains in the members of a particular gym chain over nearly 7 years.

So, one possibility is that short-term results can be extrapolated by revealing the slope of the long-term logarithmic trend. When we model long-term adaptations this way, a training intervention that increases muscle size by 10% in 2-3 months won’t continue increasing muscle size by an additional 10% every 2-3 months thereafter, but it will still lead to twice as much growth long-term as a training intervention that increases muscle size by 5% in 2-3 months. Below, you can see how that might look on both linear and logarithmic x-axes.

The second broad possibility is that gains aren’t literally logarithmic – they just slow down over time as people approach their “genetic limits.” So, when we observe a short-term difference in hypertrophy or strength gains, all we can say is that “this intervention leads to faster gains in [strength or muscle size] than some other intervention,” but both approaches would ultimately get you to the same place. The only difference is that the intervention that leads to faster gains would allow you to reach your ultimate potential a bit sooner.

The third possibility is somewhere in between the first two. Perhaps option 1 is more-or-less correct, but there’s still an eventual limit. So, for instance, let’s just assume there are three approaches to training. Approach 1 is extremely effective, approach 2 is moderately effective, and approach 3 isn’t particularly effective. Over time, you’d still achieve basically the same results with approaches 1 and 2 (you’d just achieve those results faster with approach 1), but approach 3 would still lead to sub-optimal gains over time.

Of course, long-term adaptations are likely considerably more nuanced than that in the “real world.” Whatever constitutes “optimal training” for you right now will probably change over time. But, I just wanted to point out that extrapolating the results of relatively short-term studies doesn’t require assuming that subjects will continue getting 5-10% more muscular or 10-20% stronger every 2-3 months, and I think everyone understands that perfectly well when they’re being intellectually honest.

So, personally, I do think we can extrapolate the results of volume literature beyond the 2-3 month time window of the studies on the topic. But, I also think that all three of the models above are entirely defensible. If higher volumes lead to more growth over 2-3 months, I’m pretty confident that they’ll also lead to more growth over 1-2 years (and my confidence is bolstered by the fact that the advantages of higher training volumes tend to be larger in longer studies). However, I’m entirely open to the possibility that moderate volumes and higher volumes will get you to more-or-less the same place over a period of 5-10 years. But, I am skeptical that you can maximize your long-term muscle growth with very low volumes (say, 1-3 sets per muscle group per week).

What about volume cycling?

I think it’s also worth entertaining the possibility that the volume literature provides us with evidence in favor of a volume cycling approach, rather than evidence in favor of high training volumes per se.

The idea behind volume cycling is fairly simple:

1. We adapt positively to training when a stressor is large enough to result in a stimulus for adaptation.
2. As we acclimate to a particular stressor, the stimulus it provides gets progressively smaller.
3. When we increase the magnitude of a stressor, that can cause a further adaptive stimulus.
4. The extent to which we can increase the magnitude of a stressor is finite.
5. We can significantly reduce the magnitude of a stressor while still maintaining the gains we’ve achieved.
6. We also acclimate to reductions in stress, which allows us to once again derive an adaptive stimulus from a stressor that was previously insufficient to cause adaptation.

To illustrate, let’s assume you’re totally untrained, and you start doing 5 sets of biceps training per week. At first, your biceps grow, and things are going well. However, over time, you’ll acclimate to the stressor of 5 sets of biceps training, and stop growing (or, the rate of growth will slow way, way down). But, if you increase your volume to 10 sets of biceps training per week, this increase in stress will allow your biceps to start growing again. Eventually, you plateau again, increase your volume to 15 sets, and start making gains again. However, once you reach 20 sets, you find that your biceps are struggling to recover from training. So, you reduce your volume to 5 sets of biceps training per week, which is enough to maintain the strength and size you’ve already built. After a couple of months, you can increase your volume back to 10 sets per week. Previously, this was too small of a stimulus to keep making gains, but after reacclimating yourself to only doing 5 sets of biceps training, doing 10 sets once again presents a sufficient stimulus for adaptations.

So, perhaps higher training volumes only appear to be more effective because subjects in low-volume groups may be reducing their training volumes (when they transition from self-directed training to the training they’re performing for the purposes of a study), while subjects in high volume groups are increasing their training volumes. But, if the results are primarily due to increasing training volume, rather than simply performing more training volume in a vacuum, then it’s possible that cycling through volumes between 5-20 sets per week would be just as effective as cycling through volumes between 20-35 sets per week.

I can’t let this turn into a full article on volume cycling (this article is already far too long, and I’m sure a full article on volume cycling would clock in at 4,000-5,000 words), but, broadly speaking, there are some studies that you can interpret to be indirectly supportive of the concept (one, two, three, four, five, six, seven), and a couple that provide reason to be skeptical of the concept (one, two). But, at this point, we don’t have any studies fully validating the efficacy of volume cycling.

However, we do have evidence to suggest that total training volume still matters, even in the context of increasing training volume.

The most illustrative study on the topic is a recent paper by Enes and colleagues. In this study, trained female subjects completed a 12-week lower-body training intervention with different volume assignments. One group just performed 22 sets of quad training per week. A second group started with 16 sets of quad training per week, and then increased their volume by two sets every two weeks (so they ended up performing 26 sets of quad training in the last week of the program, and averaged 21 sets of weekly quad training over the full duration of the program). Finally, a third group also started with 18 sets of quad training per week, and they increased their volume by 6 sets every two weeks (so they ended up performing 38 sets of quad training in the last week of the program, and averaged 28 sets of quad training over the full duration of the program).

The first two groups ultimately trained with similar average volumes (22 sets vs. 21 sets per week), but the group that increased their training volume by two sets every two weeks generally experienced a bit more muscle growth and gained a bit more strength. So, this provides some indirect evidence in favor of volume cycling. However, the group that increased their training volume by four sets every two weeks, and thus trained with even higher average volumes (28 sets), also experienced a bit more muscle growth and gained slightly more strength than the group that only increased their training volume by two sets every two weeks.

A prior study, also by Enes and colleagues, had similar findings with male subjects. One group increased their training volume by 4 sets per week, and another group increased their training volume by 6 sets per week (so, both groups increased their volume over the course of the training program, but the latter group ended up training with slightly higher volumes on average). The group training with slightly higher volumes ultimately gained a bit more muscle and strength than the group training with slightly lower volumes, even though both groups increased their training volumes over time.

So, I’m still fairly agnostic about the concept of volume cycling, but I do at least find it plausible that some of the observed increases in hypertrophy with higher training volumes are due to the effects of increasing training volume, rather than simply being due to the effects of high-volume training per se. However, even if some of the results of the volume literature are at least partially attributable to increases in training volume, it appears that generally training with higher volumes is still beneficial, even if you’re cycling your training volume. 

Why/how do higher volumes lead to more growth?

I think the best explanation for the superiority of higher training volumes is also the simplest one: higher volumes present your muscles with a larger total stimulus for adaptation, and larger stimuli tend to lead to larger adaptations. 

But, since it seems that people enjoy more detailed speculation (which others might erroneously refer to as “mechanistic reasoning”), I’ll indulge in a bit of it.

One speculative argument against high training volumes goes something like this:

1. The fibers of your highest-threshold motor units are your fibers with the most growth potential. 
2. Due to a combination of local fatigue and decreases in motor drive, these fibers will no longer be capable of generating sufficient tension to stimulate further increases in the adaptive stimulus they receive after a fairly small number of sets.
3. Therefore, additional sets beyond this point won’t lead to further increases in muscle growth.

However, this argument has a pretty key blind spot: you have plenty of muscle fibers that aren’t associated with your highest-threshold motor units. These fibers are more fatigue-resistant, and they’re still capable of achieving plenty of growth.

Reviews tend to find that type I fibers (which are typically associated with low-threshold motor units) experience around 60% as much growth as type II fibers (which are typically associated with high-threshold motor units) following resistance training.

So, let’s just entertain the possibility that the second assumption above is correct: with each set you perform, more of your high-threshold MUs reach the point where they’re no longer capable of generating enough tension to stimulate further adaptations. But, your remaining MUs are still capable of generating enough tension to stimulate further adaptations.

We can roughly model how this might work out in practice. For example, let’s assume that, with each set, an additional 30% of the high-threshold MUs associated with type II fibers fall below the tension threshold required to stimulate hypertrophy, and an additional 5% of the low-threshold MUs associated with type I fibers fall below this threshold. In that case, this would represent the percentage of your type I and type II fibers that can still achieve an adaptive response to each additional set.

From there, we could calculate the total hypertrophic stimulus per set, with the assumption that type I fibers experience a hypertrophic stimulus that’s about 60% as large as type II fibers during the sets where they’re producing enough tension to achieve an adaptive stimulus.

Finally, we can then calculate the total hypertrophic stimulus for the entire workout (which is just the sum of the adaptive stimuli provided by each set).

As always, I’d caution you against taking this too literally – these figures are purely for illustrative purposes. The main takeaway is just that the fibers associated with your highest-threshold motor units aren’t the only fibers capable of growth. Even after your high-threshold motor units fatigue, you still have plenty of less fatiguable fibers that will still be (relatively) unfatigued, and capable of robust hypertrophy.

Just as a general note, this conceptual model carries with it a couple of testable predictions: if this accurately describes the impact of training volume on hypertrophy by fiber type, then lower- and higher-volume training should be similarly effective for type II fiber growth, but higher-volume training should be more effective for type I fiber growth. Furthermore, this would lead to the prediction that lifters with relatively more type I fibers should attain larger benefits from high-volume training than lifters with relatively more type II fibers.

For what it’s worth, this matches my personal experience as a coach – more “explosive” lifters tended to be the ones that would benefit the most from lower training volumes, whereas less explosive lifters tended to thrive on higher volumes. I find it very plausible that the more explosive lifters had a higher proportion of type II fibers, experienced a larger relative stimulus from the first few sets they performed, and were more likely to dig themselves into a recovery hole when pushing volumes higher (since, as mentioned previously, type II fibers are more susceptible to muscle damage). I also find it very plausible that the less explosive lifters had a higher proportion of type I fibers, experienced a smaller relative stimulus from their first few sets but a larger relative stimulus from later sets (compared to the more explosive lifters), and generally experienced less fatigue and less muscle damage at any absolute level of training volume.  

Moving on, my personal pet theory is that bioenergetic adaptations are one of the primary reasons why high volumes promote larger gains in muscle mass.

As muscle fibers grow, they become less and less metabolically efficient when energy demands are high. The basic reason for this is fairly straightforward: energy demands are dictated by the contractile forces a fiber generates (each myosin power stroke requires 1 ATP molecule), and contractile proteins are more-or-less uniformly distributed throughout the muscle fiber. As a result, energy demands scale with muscle fiber volume. However, muscle fibers receive oxygen and fuel, and remove waste products via the sarcolemma (the cell membrane of a muscle fiber). As a result, a fiber’s surface area places some fundamental constraints on the oxidative capacity of the fiber.

If we assume a muscle fiber is (roughly) cylindrical in shape, then doubling the radius (or diameter) of a fiber would double its surface area, but it would quadruple its volume.

As a result, its surface area:volume ratio would decrease by 50%.

We can see the effects of this decrease in surface area to volume ratio by looking at the influence of muscle size on functional capacities. Larger muscles do tend to be stronger, but they also tend to have worse relative strength endurance.

So, as fibers grow, their peak metabolic demands increase faster than the surface area available to accomplish gas exchange, clear waste products, and bring in more fuel. In addition, diffusion distances within the fiber increase – if you’re a myofibril in the center of the fiber, and the radius of the fiber doubles, you’re now twice as far away from the sole site of gas exchange, the sole site of RNA transcription, the site where most cell signaling is initiated, the site where external energy substrates are delivered to the fiber, and the primary site of protein synthesis.

One of my long-held beliefs is that the surface area to volume ratio (henceforth referred to as “SA:V”) is the primary factor that eventually bottlenecks fiber growth. We don’t actually have much direct evidence to help us confidently determine the ultimate limiter of muscle growth, but there are a few reasons I believe the ultimate limiter is SA:V ratios (or, at minimum, something related to SA:V ratios).

The most basic reason is that SA:V ratios are a primary growth constraint in most cells that aren’t encoded with a pre-determined size (one, two, three, four, five). So, if nothing else, metabolic constraints due to SA:V ratio reductions serve as a handy null hypothesis.

The second reason is that capillary density has been found to influence hypertrophy (one, two, three). The contact points between capillaries and the sarcolemma are where gas exchange occurs. So, if the density of the actual sites of gas exchange on the cell’s surface influences hypertrophy, it’s not a big stretch to assume that surface area:volume ratios may influence hypertrophy more generally.

The third reason is that most of the “stuff” in a muscle fiber – other than the contractile proteins – is heavily concentrated near the cell membrane. That’s where you’ll find all of the nuclei, most of the proteins and metabolic machinery involved in hypertrophy signaling and protein synthesis, and the most highly active pool of mitochondria. As a result, muscle DNA scales with the surface area of the fiber, influencing transcriptional capacity, and ultimately the rate of cell growth.

The fourth reason is that negative regulators of muscle growth – namely, myostatin – appear to be sensitive to the metabolic health and oxidative capacity of the cell. Completely removing myostatin allows muscles to grow more, but this extra growth comes at the expense of the muscles’ oxidative capacity and relative force-generating capacities. Additionally, in myostatin-deficient animals, increasing the muscle’s oxidative capacity seems to be key for re-establishing normal muscle functioning. So, the functional purpose of myostatin may be to restrict muscle growth beyond the point that the muscle’s metabolic capacities can comfortably support.

So, to briefly recap: increasing SA:V ratios may be the ultimate limiter of fiber growth. Furthermore, SA:V ratios are primarily constrained in most cell types because they place constraints on the metabolic functioning of a cell.

However, metabolic functioning isn’t solely dependent on SA:V ratios (at least in eukaryotic cells). Rather, SA:V ratios are associated with metabolic functioning because they define the theoretical upper limit of gas exchange relative to the cell’s maximum rate of energy expenditure. But maximal gas exchange in skeletal muscles is ultimately dictated by capillary density, not just cell surface area. In other words, if a fiber doubled in volume, but it also doubled its contact area with surrounding capillaries, its ability to produce energy would continue scaling proportionally with its ability to consume energy (assuming mitochondrial density was also maintained).

So, my basic hypothesis is that more “metabolically fit” fibers – fibers with more capillary contact points and (potentially) better mitochondrial functioning – may be able to grow faster, and ultimately achieve a larger fiber size, since they’d be able to tolerate a lower SA:V ratio than less metabolically fit fibers.

There’s very little research investigating the limits of muscle growth, but there is plenty of evidence suggesting that the metabolic fitness of a muscle fiber influences its ability to grow following resistance training.

As mentioned previously, capillary density appears to be important for hypertrophy responses. In two separate studies in older adults, pre-training capillary density was associated with hypertrophy following resistance training (one, two). Furthermore, in younger adults, performing 6 weeks of aerobic training prior to 10 weeks of resistance training led to gains in capillary density, and substantially more hypertrophy than “just” performing 10 weeks of resistance training. In fact, decreases in capillary density (and muscle perfusion more broadly) may be one of the causes of “anabolic resistance” in older adults and adults with metabolic disorders.

Sticking with capillaries, satellite cells tend to be located near capillaries, and the satellite cells located near capillaries are more likely to be activated than satellite cells located further from capillaries. Furthermore, people with greater capillary density experience greater activation and proliferation of the satellite cell pool in response to resistance exercise and muscle damage, and training-induced increases in capillary density are associated with improved satellite cell function. Satellite cells play an important role in recovery from muscle damage, and satellite cells are the cells responsible for donating their nuclei to muscle fibers to allow for long-term growth.

Shifting our attention to mitochondria, research has found that high responders to resistance exercise display greater citrate synthase activity. Furthermore, hypertrophy is pretty strongly associated with mitochondrial respiratory capacity in middle-aged adults. Finally, other research has found that hypertrophy is inversely related to gains in mitochondrial area with resistance training, perhaps suggesting that subjects who need to enhance their local aerobic capacity experience less hypertrophy until they’ve successfully gained the mitochondria required to support larger muscle fibers.

So, I suspect that higher-volume training leads to larger gains in capillary density, mitochondrial functioning, and overall muscular metabolic capacity than lower-volume training, and that these gains in local metabolic fitness facilitate faster gains in muscle mass in the short-to-medium term, and potentially larger total gains in muscle mass long-term.

Classically, it was thought that hypertrophic adaptations and aerobic/metabolic adaptations “interfered” with each other, but more research suggests that the “interference effect” rarely needs to be a major concern. In fact, training that stimulates both hypertrophic and aerobic adaptations can sometimes lead to more hypertrophic signaling and more growth than “just” a hypertrophy stimulus. And, it’s worth noting that some of the most positive results come from studies combining resistance training and high-intensity interval training (which is, metabolically, not too dissimilar from performing several high-rep sets of resistance exercise – at least for the local musculature).

To be clear, these two possibilities aren’t mutually exclusive. Higher-volume training can present a larger hypertrophic stimulus per-workout, while also presenting a larger stimulus for angiogenesis and mitochondrial adaptations (which subsequently facilitate further hypertrophy). But, these are my two primary hypotheses to explain the effects of high-volume training that we see in the literature.

Do you really believe that strength and hypertrophy aren’t related?

This section is just to clear up a misunderstanding from the first article in this series, where I argued that, in the published literature, strength data are not particularly informative about hypertrophy outcomes.

I closed out the article with this paragraph: “Keep the context of this article in mind. I’m just discussing the relationships between hypertrophy and strength changes observed between groups and between studies in the published scientific literature. By no means am I saying that strength changes aren’t a decent indicator of hypertrophy at the level of the individual. In fact, I think that strength gains can be (and often are) a pretty good indicator of hypertrophy in certain contexts. However, those contexts significantly differ from the conditions of most studies.”

Unfortunately, it seems that some people didn’t read to the end of the article, didn’t check to see if I’d discussed the topic elsewhere (perhaps in another piece that was linked in the article, elsewhere on the site, or even in a peer-reviewed journal), or they were just in the mood to strawman my argument. I understand that the internet has a short memory, but I still found this a bit confusing. Not too long ago, there was a pretty vocal movement arguing that hypertrophy didn’t contribute to strength gains, and I was one of the people most responsible for developing and popularizing the affirmative case for hypertrophy’s contributions to strength gains – that’s why I was invited to participate in the journal article linked above (one side of a point/counterpoint; here’s the opposing viewpoint, if you’d like to check it out). I certainly wasn’t the first person to have the idea, but I was one of the first people to pull together the evidence and make a case for it.

But, I’m certainly happy to more thoroughly spell out my position here:

Within the published literature, strength outcomes and hypertrophy outcomes frequently diverge. The main reason for this divergence when comparing between meta-analyses is discussed previously in this article (strength and hypertrophy meta-analyses often use two different sets of studies that don’t fully overlap), but we also frequently see divergences within individual studies as well.

I think the main reason for the divergence (most of the time) is just that there are other contributors to strength gains beyond hypertrophy. That’s easy enough to intuit just by noticing the typical size of hypertrophy adaptations vs. strength adaptations: the average training intervention leads to a ~5% increase in muscle size, and a ~22% increase in strength. So, a little over three-quarters of the strength gains observed in most studies must be due to contributors other than hypertrophy. And, those other contributors play a predominant role in most studies, which tend to:

1. be fairly short (a few months), and
2. use folks who aren’t super well-trained

So, most of the strength gains that occur are just due to learning the lifts used to assess strength, gaining more skill with the lifts used to assess strength, or gaining experience with maxing the lifts used to assess strength. Any direct impact of hypertrophy on strength gains will typically be drowned out by those other factors. You need very large samples to actually detect a relationship between hypertrophy and strength gains in untrained lifters, and you’d need even larger samples to detect differential impacts of different training interventions on the relationship between hypertrophy and strength gains.

But, in longer studies, or in studies on lifters with higher training statuses, we DO see a stronger relationship between hypertrophy and strength gains (discussed in this article). Also, in powerlifters, we see that weight change (which is a decent proxy for hypertrophy/atrophy in that population) is the single strongest predictor of changes in strength when looking at publicly accessible large datasets (and in studies with more direct measures, the relationship is even stronger).

So, I do actually think that changes in strength can be pretty strongly reflective of hypertrophy, but only in specific circumstances:

1. If training status is already fairly high
2. If you’re assessing strength in exercises you’re already quite skilled with
3. If your approach to training remains fairly consistent (for example, if you always do sets of 8-12, and you get stronger for sets of 8-12, you’re probably gaining muscle. But, if you always do sets of 8-12, you test your 1RM, then you run a peaking block with sets of 2-5 reps and that boosts your 1RM, I don’t think that this increase in your 1RM tells you much about hypertrophy)
4. If you’re assessing strength changes over a reasonably long time scale (strength changes over one year should be more indicative of hypertrophy than strength changes over 6 months, which should be more indicative than strength changes over 3 months, which should be more indicative than strength changes over 1 month, etc.).

Those circumstances are the typical circumstances for most serious trainees, but they’re extremely atypical circumstances in most published studies. It’s very rare for a study to meet all – or even just 2 or 3 – of those criteria. Many studies use untrained or recreationally trained subjects, training interventions that meaningfully differ from the ways the subjects were training before the study, fairly short training interventions lasting around 8-12 weeks on average, and exercises the subjects may be unfamiliar with (Even if they’re familiar with the exercise at a high level, the study may require technique changes. For example, if you’re a bodybuilder who tends to squat above parallel, a study may require you to squat below parallel. You’ll get some pretty robust “strength gains” just from familiarizing yourself with below-parallel squats, regardless of the muscle growth you experience).

So, I think it’s simultaneously the case that:

1. On an individual level, strength changes can be (and often are) a pretty good indicator of hypertrophy.
2. Due to the characteristics of the studies that tend to be published, strength changes reported in the literature are typically a fairly poor indicator of hypertrophy at a group level within and between studies.

Here’s a little illustration that should help clarify how just one of these factors (study duration) could make it look like hypertrophy and strength gains are essentially unrelated, even if they happened to be perfectly related.

So, let’s assume that you start training today, and you train consistently for 20 years. Over that time, you have a great training career: you put on over 20kg of muscle mass, and you wind up nearly 3-times stronger than you started.

During your first year of training, gains in strength far outpace gains in muscle mass. But, from the start of year 2 until year 20, your strength gains perfectly mirror your gains in muscle mass.

As a result, if you could perfectly measure both strength and muscle mass, the two values would be perfectly correlated when you assessed your strength and muscularity every 3 months (starting in year 2).

However, we unfortunately can’t perfectly measure strength or muscle mass. Any method of assessing body composition comes with some unavoidable error, and you know that your strength can fluctuate a bit from day to day – some days you’re a bit stronger, and some days you’re a bit weaker. But, let’s assume that you’re exceptionally good at measuring your muscle mass and strength, such that the errors of your measurements only have a standard deviation of 1%. In other words, if you actually have 40kg of muscle mass, your measurement would come back somewhere between 39.6 and 40.4kg around 2/3rds of the time, and it would come back somewhere between 39.2 and 40.8kg 95% of the time. Or, if a bench press 10RM of 200lbs reflects your current general strength level, when you assess your bench press 10RM, you’d almost always complete 10 reps with somewhere between 195-205lbs.

Even with this small bit of measurement error included, we can still see a crystal clear relationship between hypertrophy and strength over a 20-year time span:

However, you can’t look back at a 20-year training career if you’re still in the middle of it. So, what if you wanted to see the relationship between hypertrophy and strength gains each time you assessed them (every 3 months). Surely we still see a strong correlation, right? We may not have an r² of 1.0 anymore, but you’d surely expect to still see a very clear association.

However, we very much don’t see a clear association anymore:

Keep in mind, the “true” relationship between hypertrophy and strength gains in this illustration is still a perfect, linear relationship. But, unfortunately, the inclusion of (fairly trivial) measurement errors completely obscures this association.

However, what if, instead of paying attention to the relationship between hypertrophy and strength gains over a period of 3 months, we instead paid attention to the relationship between hypertrophy and strength gains over one-year periods?

And, for the sake of argument, what if we were even more patient, and paid attention to the relationship between hypertrophy and strength gains over two-year periods?

I hope this little illustration helps clarify my point. I’m not saying that strength gains can’t be a good indicator of hypertrophy. I’m simply saying that there’s absolutely no reason to expect strength gains to be a particularly good indicator of hypertrophy in most studies that only last 2-3 months, on average. And keep in mind, this was already a very charitable illustration: hypertrophy and strength probably aren’t perfectly associated long-term, and there will certainly be plenty of other confounders in a study that weren’t present in this illustration (i.e., subjects with slightly different training statuses, overall propensities to gain muscle mass and/or strength, or pre-existing familiarity with the exercises used to assess strength). Though, I should also acknowledge that those further annoyances are somewhat counterbalanced by larger sample sizes (i.e., more than n=1), and the fact that the typical gains observed in most studies are larger than the typical gains per three months in this illustration (it’s easier to identify correlations when the data are more spread out along both axes). 

To be slightly more rigorous, here’s another illustration. I generated two datasets of 1000 subjects apiece – one for higher-volume training, and one for lower-volume training. The datasets were generated with these assumptions:

1. Lower-volume training leads to a 5±5% increase in muscle size, on average.
2. Higher-volume training leads to a 10±10% increase in muscle size, on average.
3. Hypertrophy directly and linearly increases strength gains. In other words, a 5% increase in muscle size necessarily results in strength gains that are 5% larger than they would have been in the absence of hypertrophy.
4. The other factors that influence strength (“neural adaptations,” intramuscular connective tissue adaptations, improved familiarity with strength assessments, etc.) additively and independently increase strength gains by an additional 15±15% on top of the strength gains directly attributable to hypertrophy. 

To briefly justify these assumptions:

1. For assumptions 1 and 2, a 5% difference in hypertrophy is in line with the magnitude of difference we should expect with a reasonably large difference in training volumes (around 20 sets per week), according to the Pelland meta-regression (the linear marginal effect per set was 0.24%. That value is a bit larger when doing fewer sets overall, and a bit smaller when doing more. See Figure 5B).
2. For assumption 1, 2, and 4, hypertrophy and strength change scores tend to have coefficients of variation (mean change divided by standard deviation of the change) that remain remarkably close to 1.0. See Figure 8 here.
3. For assumption 4, as mentioned previously, relative strength gains tend to be around 15% larger than relative increases in muscle size, on average.
4. Assumption 3 is what we’re ultimately observing the effect of (i.e., we’re just taking it to be true, and observing the implications of that assumption). In other words, we want to know what sorts of results we should expect to see if hypertrophy does directly, linearly increase strength gains.

So, here are the resulting datasets:

These associations between hypertrophy and strength gains (r ~ 0.25-0.60) are in line with what we should expect to see, based on other research on the topic. The correlation between hypertrophy and strength gains is a bit stronger in subjects with high training status, and a bit lower in untrained subjects, but these associations match the typical associations we tend to see in subjects with low-to-moderate training status (i.e., the typical subjects in most studies).

From here, we can simulate the results of “studies” performed on these subjects. If we randomly selected groups of 20 subjects from both of these populations, we would generally observe more hypertrophy and larger strength gains in the groups training with higher volumes in most studies. The average difference (in favor of higher volumes) would be around 5±2.5% for hypertrophy, and 5±5% for strength. Since the between-groups differences in strength outcomes are inherently more variable, we’d expect these differences to be statistically significant in about 50% of studies for hypertrophy outcomes, but only in about 17% of studies for strength outcomes. As a result, the plurality of studies (about 46%) wouldn’t find that higher volumes significantly increase hypertrophy or strength outcomes. Furthermore, studies finding that higher volumes significantly increase hypertrophy but not strength would be expected to outnumber studies finding a significant impact on both hypertrophy and strength with a ratio of about 2.5-to-1 (37% vs. 14%).

If we instead only had 10 subjects per group (not uncommon for the hypertrophy literature), we’d expect to see similar patterns, but way fewer statistically significant results overall: we should only expect to see statistically significant results for hypertrophy around 30% of the time, and statistically significant results for strength gains around 10% of the time. Furthermore, we should observe nominally more hypertrophy with lower training volumes in around 8% of studies, and nominally larger strength gains with lower training volumes in nearly 30% of studies. As a result, most studies shouldn’t find a significant impact of training volume on hypertrophy or strength outcomes (around 65%), and when a significant effect is observed, studies that only find a significant effect for hypertrophy should be expected to outnumber studies that find a significant effect for both hypertrophy and strength by a nearly 4-to-1 ratio (24% vs. 6.5%). 

I’d just like to reiterate that, for the purposes of this illustration, we know that higher volumes lead to more hypertrophy than lower volumes, we know that hypertrophy directly and linearly increases strength gains, and so we know that higher volumes also lead to larger strength gains than lower volumes. So, even if all of those assumptions are true, we should still expect to see plenty of studies where it appears that higher volumes don’t lead to significantly more hypertrophy, significantly larger strength gains, or both. Furthermore, studies that appear to have divergent outcomes (higher volumes lead to significantly more hypertrophy but not significantly larger strength gains) should be expected to outnumber studies that have convergent positive outcomes (significantly more hypertrophy and significantly larger strength gains) by a roughly 3-to-1 ratio.

As a general note, this is one of the biggest reasons why meta-science (systematic reviews, meta-analyses, and meta-regressions) is so valuable. Even when people pay lip service to the idea that “you shouldn’t put too much faith in the results of a single study,” I think there’s still a tendency to underestimate just how variable individual studies’ results can be. In the illustration above, all of that variability in results is purely due to random sampling variance. When you add in effects from other factors that could influence results (different subject characteristics, different training protocols, different study durations, different techniques used to assess the outcomes of interest, etc.), you should expect an even larger spread of findings. But, when you can pool the results of many studies, true underlying effects become much clearer. 

As one final note, within the context of your day-to-day training, strength gains can be a useful proxy for hypertrophy (with all of the caveats listed above). However, within the context of the scientific literature, I think it’s far more justifiable to make (tentative) inferences about strength from hypertrophy data than it is to make inferences about hypertrophy from strength data.

Going all the way back to the 1970s, Moritani and deVries proposed that strength gains that occur early in a training intervention are primarily due to neural adaptations, with hypertrophy playing an increasingly important role over time. This view was solidified in an influential review paper by Sale and colleagues from 1988. This has been the dominant view ever since. The “crossover point” – the point at which strength gains switch from being primarily the result of neural adaptations to being primarily the result of hypertrophy – likely occurs around 8-12 weeks into a training program. And, incidentally, most training studies only last about 8-12 weeks.

As a result, most studies simply aren’t designed to capture the impact of hypertrophy on strength gains. Most of the strength gains that occur in most studies are primarily attributable to “neural adaptations,” and simply practicing the exercises used to assess strength. 

Previously, I noted that making inferences about hypertrophy from strength data can lead you to some pretty unjustifiable conclusions: namely, that you can maximize hypertrophy by doing just 5 sets per week with 10 reps in reserve. However, the same is also true if you make (long-term) inferences about strength changes from the strength data. I don’t think anyone actually believes that you can maximize long-term muscle growth by doing 5 sets per muscle group per week with 10 RIR, but I also don’t think anyone actually believes that you can maximize long-term strength gains by doing 5 sets per exercise per week with 10 RIR. However, within the context of typical 8-12 week training studies, we do observe that strength gains can be maximized with very few sets (particularly in studies with untrained subjects), and with very high RIRs.

In other words, to bring this article full-circle, I don’t think strength data is particularly informative about hypertrophy, but I do actually think that hypertrophy data can be informative about how to train for long-term strength development.

So, when strength and hypertrophy results diverge, I think it’s entirely justifiable to use the hypertrophy results to shade your interpretation of the strength findings. For example, “over 8-12 weeks, we see that you can maximize strength gains when training with 5+ reps in reserve. However, training closer to failure is important for maximizing hypertrophy. So, in the long run, you likely need to do at least some of your training considerably closer to failure to maximize strength gains, because muscle size will eventually become a limiting factor for strength performance.” Or, “over 8-12 weeks, we see that untrained lifters can maximize strength gains with quite low training volumes. However, higher training volumes are necessary to maximize hypertrophy. So, in the long run, you likely need to train with somewhat higher volumes to maximize strength gains, because muscle size will eventually become a limiting factor for strength performance.”

What should my volume be per-workout?

To this point in the article, I’ve almost exclusively focused on weekly volume. But, you’re likely wondering how that volume should be split up over the course of a training week.

In the time since I started writing this article, a new meta-regression was published (by Remmert and colleagues) analyzing the relationship between hypertrophy and per workout training volume. 

Ultimately, its results are pretty well-aligned with the Pelland meta-regression on weekly training volume. It also found that muscle growth generally tended to increase and per-workout training volume increased:

For what it’s worth, the researchers calculated the “point of undetectable outcome superiority” (PUOS) in this meta-regression, which is basically the point at which the data suggests that you should be less than 50% confident that further increases in volume will yield further increases in strength. The PUOS was 11 fractional sets (with fractional sets, you’d count each set as 1 set for the primary muscle(s) trained by an exercise, and as 0.5 sets for secondary muscles).

So, boring answer: 11 sets

But, as we’ve already discussed in this article, your optimal level of volume may be lower (or higher) than 11 sets per workout. Furthermore, the optimal per-workout volume must interact with training frequency to some degree: I wouldn’t be surprised if you could benefit from more than 11 sets if you just trained a muscle group once per week, but I think it would be unwise to train a muscle group with 11 sets, 6 days per week.

If you want to give reasonably high-volume training a shot, I think 20-30 weekly sets is a pretty good starting point. Just divide those weekly sets by your desired training frequency, and you’ll have your answer. If you want to do around 25 sets, you could do 5 sets on 5 days, around 6 sets on 4 days, 8 sets on 3 days, or 12 sets on two days. 

What about diminishing returns, efficiency, and “junk volume”?

One popular argument against higher training volumes is that they’re less efficient than lower training volumes, since diminishing returns kick in as volume increases.

On one hand, this is 100% true: we do see diminishing returns as volume increases.

On the other hand, this is 100% unrelated to the topic at hand: how much volume is required to maximize muscle growth?

Diminishing returns aren’t lower total returns. “Diminishing returns” just mean that the marginal utility of each additional set is smaller than the marginal utility of the previous set. In other words, if you could increase your muscle size by 3% with a weekly volume of 6 sets, or 6% with a weekly volume of 18 sets (which is roughly what’s implied by the Pelland meta-regression), you’re seeing diminishing returns in action. In order to double your rate of muscle growth, you need to triple your training volume. So, the average marginal utility of sets 1-6 (around a 0.5% increase in hypertrophy per set) is approximately twice that of sets 7-18 (around a 0.25% increase in hypertrophy per set).

So, technically speaking, diminishing returns kick in immediately. Set 2 nets less additional growth than set 1, set 3 nets less additional growth than set 2, set 4 nets less additional growth than set 3, etc.

So, you might reasonably look at that and decide that the marginal benefits of each additional set drop off enough that it doesn’t make much sense to do more than 10 sets per week. We can scale the graph above differently to drive this point home, by showing the value of each set relative to the value of your first set of the week.

Even by set 5, the additional growth you achieve with each additional set has already been halved. At 15 sets per week, it takes three additional sets in order to achieve the additional marginal value of your first weekly set. We definitely see a decrease in training efficiency with each additional set we perform.

However, diminishing marginal gains are still marginal gains. The amount of additional hypertrophy you achieve per set decreases, but the total amount of hypertrophy you achieve is still increasing as volumes increase. So, for instance, if you determined that it wasn’t worth your while to do more than 10 sets per week due to the progressive decreases in training efficiency, you might only spend 1/3rd as much time in the gym as someone aiming to maximize their muscle growth with ~30 sets per week, but you might also be leaving approximately half of your potential gains on the table.

As with most things in life, achieving efficient results and achieving maximal results are mutually contradictory goals. If your primary goal is maximizing efficiency, you should train with reasonably low volumes. But, the decrease in efficiency with higher volumes does not mean that higher volumes can’t still offer a lot of value for someone with a primary goal of maximizing total muscle growth.

This is obviously a fairly rough ratio (since there’s still a reasonably wide confidence interval around the main meta-regression line in the Pelland paper), but it seems like volume and hypertrophy scale with a roughly 3:2 ratio – training with 3-times higher volumes yields approximately twice as much growth, at least through the volume ranges for which we have a reasonable amount of data.

To be entirely clear, I’m not trying to encourage people to train with higher (or lower volumes). I simply want people to be informed about the implications of their choices. If you’re happy with the results you’re achieving with 5 sets per week, and you don’t like the idea of spending three times as long in the gym, I absolutely would not recommend increasing your volume to 15 sets per week. I encourage you to think through the tradeoffs for yourself.

To quickly wrap up this section, I’d just like to briefly touch on the concept of “junk volume.” Junk volume is an ill-defined term – sometimes people use it to mean “additional volume that yields diminished returns,” and sometimes people use it to mean, “additional volume that yields no additional returns.”

As we’ve seen, there are diminishing returns with higher training volumes, but one person’s trash (or in this case, “junk”) could be another person’s treasure. You may not like the idea of doing additional sets that only have 25-50% as much value as your first set of the week, but plenty of people would gladly take that deal.

Furthermore, as mentioned previously, we really don’t know the absolute maximum volume that people can benefit from (refer to the FAQ titled: “Where’s the actual limit?”). It may be the case that doing more than 25 sets is “junk volume” (i.e., additional volume that yields no additional growth). Or, it’s entirely possible that “junk volume” doesn’t kick in until you’re doing more than 50-60 sets per week. I do think there’s a limit somewhere, but anyone who claims to know where it is (with any reasonable degree of confidence) is either uninformed, or they’re lying to you.

However, the scenario where I do think additional volume can start becoming pretty “junky” is the one discussed in the FAQ titled “Why didn’t higher training volumes work for me?” It’s not terribly uncommon for lifters to significantly dial back their per-set effort when training volumes increase. And, I think this is particularly common near the end of long workouts. For example, let’s assume that you’re doing 10 sets of quad training, three times per week. In each workout, you do 3 sets of squats, 3 sets of leg press, 2 sets of lunges, and 2 sets of knee extensions. If you do your squats and leg presses with 0-2 reps in reserve, but you seriously half-ass your lunges and knee extensions, I think it’s quite plausible that those half-assed sets of lunges and knee extensions are “junk volume” that isn’t going to improve your results. But, I don’t think that necessarily means that doing 4 more sets of quad training after your squats and leg presses would inherently mean you’re doing 4 sets of “junk volume.” If you put a high degree of effort into those 4 sets, I think they’d still have value. But, if you don’t, then I doubt that those last 4 sets will be doing much for you.

How much volume do you actually recommend?

This article is coming to a close, so now it’s finally time to get practical. After so much discussion of training volume, how much training volume do I actually recommend?

For starters, I think this is a topic of purely academic curiosity for most people. For the vast majority of folks, your training volume is more-or-less dictated by your life outside of the gym. Between work, social obligations, commuting, etc., you might have 2-4 hours per week in the gym, which won’t give you enough time to bomb every body part with 40 sets per week. For most people, “high-volume” training just means using your limited gym time efficiently, not spending 30 minutes warming up, not resting for 5 minutes or getting distracted by social media between sets, and getting in as much high-quality training as your limited time allows (which will probably still come out to <15 sets per muscle group per week).

But, if you don’t have (major) time constraints on your training, my training advice is extremely simple:

If you’re currently making solid, consistent gains training however you currently train, don’t change a thing. A bird in the hand is worth two in the bush. Doing something else with your training may produce faster results in theory, but you already know what’s producing solid results in practice. 

But, if you’ve plateaued, it may be time to do some theorycrafting and troubleshooting.

In that situation, you have two options:

1. If there’s a style of training that’s previously worked pretty well for you, and it’s reasonably dissimilar to how you’re training now, I’d typically advise giving it another shot (perhaps verbatim, or perhaps with some tweaks based on things you’ve learned about your body since you last trained in that style).
2. If option 1 doesn’t apply to you, or if you just think it sounds boring (no judgment), then it’s time to try something new.

“Trying something new” can typically go in one of two directions:

1. Iterate on your current program (keep the bones intact, and make some fairly small tweaks).
2. Throw caution to the wind and try out an entirely new style of training.

I know that many people firmly recommend the first option, but I think both are perfectly valid. On one hand, if your current program did work pretty well for you for quite a while, iterating on it is the safe option. However, it’s also entirely possible that the best style of training for you looks very different from your current program, and it might take 10 years to find it if you’re just cautiously making small tweaks the whole time.

If you go with option 1 (iteration), there are any number of ways you could tweak your program: you could swap out exercises, shift to a slightly higher or lower rep range, increase or decrease your frequency, etc. And, of course, increasing volume is one extremely valid option – since you made it to this point in the article, I’ll assume that increasing volume is something that you’re interested in trying.

If you go with option 2 (tear it all down), my genuine advice is to seek out a program or style of training that looks fun to you. I’ve found that, more often than not, the style of training you enjoy doing also tends to be quite effective training (as long as it’s sufficiently challenging and structured in some way – there are limits, of course). This could be physiological (i.e., your body gives you some sort of positive feedback when you’re training in a way that suits it, and negative feedback if you’re training in a way that doesn’t suit your physiology), but I suspect it’s primarily just a matter of people training harder and more consistently when they enjoy their workouts.

But, for the sake of discussion, I’ll just assume that high-volume training sounds fun to you if you’ve made it to this point in the article.

So now let’s zero in on the topic at hand: if you do want to increase your training volume, or you just want to give “high volume” training a shot for the first time, how should you go about it?

You have four basic options, but I’d only recommend three of them. I’ll list them in order of “what I’d least recommend” to “what I’d most recommend.”

Option 1: Dive in head first

This is the one option I’d caution against. I don’t think it’s “bad,” and I think fearmongering about high-volume training is seriously overblown (“you’re definitely overtraining,” or “you’re definitely going to get injured”). There are just three other options that will allow you to arrive at the same destination, potentially mitigate risk a bit, and learn a bit more about yourself along the way.

But, if you do go this route, I’d recommend starting on the low end of “high-volume”: either increase training volume on your current program, or write (or find) a new training program that will situate you with around 20-25 sets per body part per week.

When selecting exercises, be mindful of choosing lifts that don’t have a tendency to cause joint or connective tissue discomfort for you. For example, I’d personally be cautious with overhead presses (for myself – this isn’t generalized advice) since they have a tendency to cause me some shoulder discomfort, and I’d instead prioritize various delt raises. In your case, you may have a history of biceps tendonitis that flares up when you do a lot of preacher curls, and decide it may be preferable to prioritize cable curls or concentration curls instead. Across the board, I’d strongly caution against doing high-volume squats and deadlifts simultaneously – you don’t need to avoid them altogether, but it’s probably wise to primarily increase your lower-body volume with lifts that don’t load the spine quite as heavily: leg press, split squats, knee extensions, leg curls, back raises, hip thrusts, etc.

I’d also generally recommend training most muscle groups 2-3 days per week. If you only train a muscle once per week, I suspect that hammering a muscle with 20-25 sets in a single workout does probably exceed the point at which more volume in that session ceases to be productive. With even higher frequencies (say, 4-5 days per week), each individual workout will tend to be a bit easier. However, I find that, when training with higher volumes, it helps to still have at least one period each week where you can rest a muscle for at least two days in a row, which is much easier to accommodate with a moderate frequency. 

It also wouldn’t be a bad idea to give yourself a week or two to acclimate to the workload before ramping up your per-set intensity. In other words, if you’re jumping from 10-15 sets per week to 20-25 sets per week, take a week to train with around 3-4 reps in reserve on most sets. In the second week, maybe aim to keep 3-4 reps in reserve for about half of your sets, and push closer to failure on the other half. From week 3 onward, keep most sets quite challenging – generally train within about 2 reps from failure for most compound lifts, and push to failure for most sets of single-joint exercises.

Keep tabs on your body as you’re acclimating to the increased workload. Muscle soreness, aches, and general fatigue are perfectly normal. Joint or tendon pain (especially pain that increases under loading) is not. If you’re having issues, you may need to sub out exercises, or even dial the volume back down for a particular muscle group. However, the typical experience is just a lot of soreness and fatigue for about 3-4 weeks. Past that point, you acclimate to the increased workout, soreness and fatigue begin to dissipate, and you can settle into a good groove.

Finally, the most important bit: keep the purpose of the volume increase in mind. The goal isn’t just to use more volume for the sake of using more volume – you’re training with more volume for the purpose of improving your results. So, while you’re in the process of acclimating to the increase in volume, it’s very common for your training loads to trend down a bit. However, within about 2-3 months, your training loads should exceed the loads you were using before you increased your training volume. If so, great! That’s an excellent indication that this has been a successful experiment, and you likely respond pretty well to higher training volumes. If your training loads are similar to what they were before, that’s a fairly neutral signal. On one hand, being able to do more sets with the same loads for the same number of reps does indicate an increase in performance, but on the other hand, this increase in performance may just be due to an increase in strength endurance. So, if you’re simply enjoying training with higher volumes, you can stick with it, but you’d also have every right to give something else a shot instead. However, after 2-3 months of training with higher volumes, if your training loads are still lower than they were before you increased your volume, that’s a pretty strong indication that you should give something else a shot.

If you are responding well to your new high-volume program, we’re back to square 1: you’re running a program that’s producing good results, so don’t change anything. But, when you do plateau again, this experiment gives you a pretty good indication that further increases in training volume may be worth trying. You responded well to 20-25 sets, so it may be worth giving a volume of 25-30 sets a shot. Or, if you want to try something else, this is still a valuable experience that could inform program adjustments in the future (i.e., if you switch back to a lower volume approach, you can have a bit more confidence that you’d respond well to increasing your volume once you plateau again).

Also, I feel like this should go without saying, but don’t try to significantly increase your training volume while simultaneously trying to lose weight. Save volume increases for a period of time when you’re trying to gain (or at least maintain) body weight.

Option 2: Progressive volume cycling

With this approach, instead of increasing your volume all at once, you’d increase it in waves.

For example, let’s just assume you currently do 8 sets per muscle group per week. Instead of just jumping straight to 20 sets, you could do something like this:

Week 1: 8 sets

Week 2: 12 sets

Week 3: 16 sets

Week 4: 20 sets

Week 5: 10 sets

Week 6: 14 sets

Week 7: 18 sets

Week 8: 22 sets

Week 9: 12 sets

Week 10: 16 sets

Week 11 onward: 20 sets

This approach allows you to acclimate to higher levels of volume as you go, but gives you a week here and there to recuperate before building back up. It also gives you an opportunity to experience training with 20+ sets twice before you finally settle at that level of volume for a more extended period of time. So, it should help mitigate some of the early initial fatigue you’d otherwise deal with by jumping straight to 20 sets per week.

All of the same recommendations I gave for Option 1 still apply (be mindful of exercise selection, don’t try to do this while you’re cutting, etc.). However, I just want to reiterate that you need to keep the purpose in mind: you’re aiming to increase your training volume for the purpose of achieving better results, not just for the sake of increasing your training volume. 

Ultimately, I think this is a perfectly fine approach. I have no principled objection to it, really. I just think it’s more complicated than it needs to be, and that option 3 will get you to the same place in a much more straightforward manner.

Option 3: The gradual ramp up

With this approach, you’re just going to increase your volume by about 20% every 2-4 weeks. 

So, if you’re doing 10 sets per body part per week now, bump it up to 12 sets. If that’s feeling fine after 2 weeks, go up again to 14 or 15 sets. If you still feel a little fatigued after two weeks, stick with 12 sets for another week or two before going up. From there, just repeat the process until you get up to 20-25 sets per week. 

Keep in mind, I’m writing this section with the assumption that you’re aiming to try “high volume” training (which we’re operationally defining as training with 20+ sets per muscle group per week). If that’s not the case, what I’d actually recommend is just increasing your training volume by around 20%, sticking with that volume for a few months, and monitoring your progress. If you’re making better gains than you were previously, great! Stick with it. And, since an increase in volume worked well this time around, that’s a solid indication that another 20% bump in volume may get things moving again the next time you stall. However, if that bump in volume doesn’t move the needle, that’s an indication that you probably shouldn’t get too tunnel-visoned when it comes to further volume increases. You could test out an additional bump in volume, but it may be worth giving another strategy a shot instead.

Option 4: Body part specialization cycles

The main reason I tend to favor body part specialization blocks is that they let you take a test drive of high volumes without having to fully commit to the strategy.

The idea is pretty simple. Pick one or two muscle groups you’d like to target. Ramp up the volume for those 1-3 muscle groups, but stick with a “normal” level of volume for the rest of your muscle groups. This lets you avoid the risk of building up generalized fatigue, while still seeing if you respond well to higher training volumes.

If it goes well, you could just stick with a strategy of running rotating body part specialization cycles. But, you could also use this as a strategy for gradually ratcheting up your “normal” training volume. For example, if you currently do 10 sets per muscle group, you could start by running a 2-month specialization cycle for your biceps and triceps, targeting them with around 20 sets per week. After this cycle, you can drop the volume for your biceps and triceps back to 15 sets, which should now feel relatively easy. From there, run a specialization block for another 1-3 muscle groups. Repeat the process until you’ve established a new “baseline” volume level at 15 sets. After that, you could just repeat the process again, running specialization blocks with 25-30 sets, then settling back to a volume of 20 sets per muscle group.

Specialization blocks can help you avoid the common tendency of sandbagging on some of your sets as volume increases. You know the whole point of a specialization phase is to really target your selected muscle groups, which can help you keep a focus on training them with a high level of effort (instead of potentially going a bit easier on some sets, because you feel like you need to pace yourself for a long workout with a bunch of sets for a bunch of different muscles). So, I think they give you the best odds of success. If one of the other strategies listed above doesn’t “work” for you, that may mean you don’t respond positively to higher training volumes, but it may also mean that you mentally struggle to put the necessary level of effort into each set when you increase your training volume. So, specialization blocks provide you with the clearest signal you can hope to get in a reasonable amount of time. If the muscles you target grow noticeably faster during the specialization block, that’s a very clear signal that you can benefit from higher training volumes – it’s now just a matter of determining how much you can/should increase your volume, and how quickly you should do it. However, if you run a couple of specialization blocks and come away from them with nothing to show for it, that’s a very clear signal that increasing your training volume probably isn’t the most productive way for you to stimulate more growth.

So, to finally answer the question posed at the start of this section (“how much training volume do I actually recommend?”): I think you should do the amount of volume that you need to do. If you’re making great progress with your current level of training volume, stick with your current level of training volume. When you stall, increasing volume is one of the strategies you could try to get progress moving again. At some point, I would recommend that you give higher-volume training a shot, especially if you’ve mostly trained with fairly low volumes (<10 sets per muscle group per week) to this point. I think around one-fourth of people will find that they get dramatically better results with higher volumes (with a big enough difference in results to easily justify the additional required time investment), around one-fourth of people will find they get somewhat better results with higher volumes (a noticeable difference, but a small enough difference that the additional time cost may not be “worth it” unless you’ve very serious about maximizing your results), around one-fourth of people will find that their results aren’t really affected positively or negatively, and around one-fourth of people will find that they actually get a bit better results with lower volumes. But, across the board, the goal should be to find the level of volume required to help you achieve the results you want to achieve; the goal is not to constantly increase your training volume, train with the highest level of volume you can recover from, or reach some arbitrary volume target on the assumption that doing so will give you the results you want to achieve.

On average, I do think hypertrophy is maximized with pretty high training volumes, but I also think the optimal level of volume for individuals can vary considerably. At the moment, the research doesn’t tell us “the” average level of training volume required to maximize muscle growth. But, even if it did, I still don’t think it would be wise to treat that single value as the level of training volume that everyone should aim for.

I think endurance training offers a useful point of comparison, because there’s a lot more sports science research investigating and documenting the training practices of elite endurance athletes than elite strength athletes or bodybuilders. Obviously there’s more that goes into endurance training than total mileage, but almost all elite runners put in at least 100km (around 60 miles) per week, and training volumes pushing 200km (around 120 miles) per week aren’t unheard of. I don’t think endurance athletes (and researchers studying endurance athletes) are preoccupied with identifying a single “optimal” level of training volume to maximize endurance performance, but if such a value exists, it’s very unlikely to be below 100km per week.

So, if someone asked you how much they should run each week to improve their aerobic fitness, would you tell them that they need to be running at least 100km per week?

I certainly hope not. I’d hope that you’d meet them where they were at, evaluate their current training program, and ask some questions about their goals (are you trying to qualify for the Olympics, or are you just hoping to get your 5k time below 30 minutes?) and lifestyle (do you have time to pound pavement for 6-10 hours per week, or do you have a busy life with numerous responsibilities that impose time constraints?) before making any concrete recommendations.

That’s more-or-less my take on training volume for muscle growth. I do think that, on average, the average volume required to maximize muscle growth is probably at least 20-25 sets per week (and potentially higher), but I think the volume required to maximize hypertrophy may be quite a bit higher or lower on an individual basis. And, even if the volume required to maximize muscle growth for you is, say, 40 sets per muscle group per week, that still would not necessarily imply that you should do 40 sets per muscle group per week (“is” does not necessarily imply “ought”). It may not comport with your schedule and lifestyle (i.e., you may only have 2 hours per week to spend in the gym), and it may not be necessary for your goals. If you want to compete in the Mr. Olympia competition, you probably do need to train in a manner that would absolutely maximize hypertrophy, but if you just want to look swole at the beach, or put on some mass to be more competitive in some other sport, you probably don’t need to train in a manner that would absolutely maximize hypertrophy. 

So, the simplest answer to the question posed by this section header (“How much do you actually recommend?”) is that I think you should do the amount of volume that’s required to achieve your desired results. That may be 5 sets per week, or it may be 30. If you’re aiming to build a solid amount of muscle without needing to spend too long in the gym, I think ~10 sets per week is generally a pretty good starting point. If you’re aiming to absolutely maximize hypertrophy, I think ~15-25 sets is generally a pretty good starting point. From there, it’s just a matter of troubleshooting and self-experimentation to determine the level of training volume you respond best to (however you define “respond best” – that could mean absolutely maximizing results, or it could mean achieving pretty solid outcomes with minimal time in the gym). 

I realize that may be an anticlimatic way to wrap up this article, but it’s the message I want to leave people with, and I’m tired of writing now.