Muscle Hypertrophy

On this page

• How many sets are you actually doing?
• How many sets per muscle per week?
• How close to failure should you train?
• What do you have to eat to grow?
• Does training frequency matter?
• Why are you not growing?
• Does this change for women, older lifters, or trained lifters?
• What is still genuinely contested?
• What do the drugs and supplements actually do?
• Sources

Almost everyone miscounts their training volume, and volume is the variable with the clearest dose-response. The sets you write in a log are not the sets your muscles receive. Count them properly and most programming problems become obvious.

How many sets are you actually doing?

The sets you write in a log are not the sets your muscles receive. Enter your weekly hard sets per exercise and the auditor distributes them by fractional counting, showing where each group lands against the dose-response evidence.

How the counting works

Direct sets count as 1.0, indirect sets as 0.5. That 0.5 is a convention, not a measured constant. The underlying meta-regression established that fractional counting predicts outcomes better than counting direct sets alone; it did not fix the exact weight, and reasonable people use 0.3 to 0.5.

Exercise-to-muscle assignments below are also a judgement. Check them against your own execution, since a row rowed with a big arm bend trains biceps more than one rowed with the elbows locked in.

How many sets per muscle per week?

Ten to twenty fractional sets per muscle per week is the best-supported range. Under four is below where growth is reliably observed; past twenty, returns flatten while fatigue, joint load and time cost keep rising. This is the best-established relationship in hypertrophy training, and it is a curve with diminishing returns rather than a threshold.

The older estimate of roughly +0.37% muscle size per additional weekly set came from a 2017 meta-analysis. The 2025 meta-regression of 67 studies and 2,058 participants confirms the direction while showing the curve bending: each additional set buys less than the one before it.

What this is not

These bands describe averages across studies. Individual response varies enough that the honest use of a number like 12 is as a starting point you then adjust from, judged on whether you are actually growing.

How close to failure should you train?

Zero to three reps in reserve is the working range for most sets. A set only counts as a hard set, and training closer to failure produces slightly more hypertrophy — an effect smaller than gym culture assumes.

Meta-regression across 15 studies found a small advantage for sets taken closer to failure, effect sizes around 0.15–0.21. For strength the difference was negligible.

• 0–3 reps in reserve is the working range for most sets.
• 4 or more reps in reserve contributes materially less, and sets that far short should probably not be counted as hard sets in the auditor above.
• Failure on every set buys little extra and costs recovery, particularly on heavy compounds.

The practical consequence is that people who train far from failure need more volume to get the same result, and people who take everything to failure usually need less.

What do you have to eat to grow?

Total energy first, then total protein at 1.6 g/kg/day, then sleep. Training supplies the signal; nutrition decides whether the signal can be acted on, and it is the variable most often responsible when a well-designed program produces nothing.

Energy

Muscle is expensive to build. In a deficit you can still gain, but the conditions narrow sharply: it happens most reliably in the untrained, the detrained returning after a layoff, those carrying more body fat, and those coming back from a period of under-eating. A trained lifter at 12% body fat in a 500 kcal deficit is not in that group.

Larger surpluses do not build muscle faster. They add fat faster, and the muscle gain plateaus well before the fat gain does. The rate-of-loss side has the cleanest data: elite athletes losing at 0.7% of body weight per week gained 2.1% lean mass over the intervention, while those losing at 1.4% per week gained none, with comparable fat loss between groups.

Protein

• 1.6 g/kg/day is where the meta-analytic benefit plateaus, from 49 studies and 1,863 participants. The confidence interval extends to roughly 2.2 g/kg, which is the defensible upper bound when lean or in a deficit.
• Total daily intake dominates. Distribution matters far less than the total, and the effect of protein supplementation over and above adequate food is small, roughly a third of a kilogram of extra fat-free mass across a training study.
• Four feedings of about 0.4 g/kg is a sensible default rather than a requirement. The differences between distribution patterns in controlled work are small.
• The anabolic window is hours, not minutes. Muscle protein synthesis stays elevated for 24 hours or more after training in trained lifters. Eating within a fixed post-workout window has not held up as an independent variable once daily intake is matched.
• Protein quality matters at the margins. Leucine content drives the acute response, which is why plant sources generally need a somewhat higher total intake to match animal sources.

Carbohydrate and fat

Neither is directly anabolic in the way protein is. Both matter for what they let you do.

• Carbohydrate fuels the training that produces the stimulus. Low availability reduces work capacity within a session and across a week, so its effect on growth runs through volume. It also supports the hormonal environment: carbohydrate restriction is one of the routes by which under-eating suppresses the gonadal axis.
• Fat below roughly 0.5–0.8 g/kg/day starts to interfere with hormone production. Very low fat intakes modestly lower testosterone.
• Alcohol impairs muscle protein synthesis acutely, and the effect persists even when protein is co-ingested. It also degrades sleep quality, which compounds the cost.

Sleep

One week of five-hour nights lowered daytime testosterone by 10–15% in healthy young men. The more immediate cost is to training quality: fewer reps at a given load, worse effort regulation, and slower recovery between sessions. Sleep is usually the highest-yield correction available, and the one people are least willing to make.

The order that matters

Total energy, then total protein, then sleep, then everything else. The remaining variables are real but small, and pursuing them while any of the first three is unresolved is how people spend years optimising the wrong thing.

Does training frequency matter?

Once weekly volume is held constant, spreading it across more sessions adds little or nothing to hypertrophy — frequency buys set quality, not an extra growth signal. It is the most over-discussed variable in training. The 2025 meta-regression found frequency had little independent effect on muscle size, though higher frequencies modestly favoured strength with the same diminishing returns.

That finding contradicts a lot of popular programming advice, and it is worth being precise about what it does and does not mean.

What frequency is actually for

• Distributing volume so the sets are better sets. Sixteen sets for a muscle in one session and sixteen across two are not equivalent in practice, because the last sets of a very long session are performed fatigued and closer to failure than intended. Frequency buys set quality, not an extra growth signal.
• Fitting the week you actually have. A schedule you complete beats a better schedule you skip.
• Managing per-session ceilings. There is reasonable evidence for diminishing returns within a single session for a given muscle, which is the practical argument for two sessions rather than one.

What it is not for

The common argument runs from muscle protein synthesis: MPS is elevated for roughly 24 to 48 hours after training, so training a muscle more often keeps it elevated more of the time. That mechanism is real, and it has not translated into a frequency advantage in volume-matched outcome studies. Acute MPS measurements are a poor predictor of long-run hypertrophy, which is a useful general caution about mechanism-led training arguments.

Choose frequency to make your sets better. Do not expect it to do anything on its own.

Why are you not growing?

Work this in order. Most people jump to the bottom of the list and change their program when the answer is at the top.

1. Energy. Are you eating enough to support growth? Chronic under-eating is the most common cause and the one people least suspect, because it does not feel like a training problem.
2. Volume. Run the auditor. Most stalls in a specific muscle are that muscle sitting under 6 fractional sets while its neighbours sit at 15.
3. Effort. How many reps are actually left? Self-reported RIR is consistently overestimated, meaning people are further from failure than they think.
4. Progression. Is load or reps documented as going up over months? Doing the same sets with the same weight is maintenance, however hard it feels.
5. Measurement. Are you actually measuring? Muscle gain in a trained lifter runs a few hundred grams a month, which is invisible in the mirror and below the noise on a bathroom scale.
6. Recovery. Sleep, life stress, and endurance volume high enough to interfere.
7. Time. Trained lifters gain slowly. A plateau over six weeks is often not a plateau.
8. Medical. If you are doing the above and losing muscle, strength or drive, that is worth investigating rather than programming around.

Does this change for women, older lifters, or trained lifters?

Most of the hypertrophy literature is built on young, mostly male, mostly untrained or recreationally trained participants in 8 to 12 week studies. That is the population the numbers on this page describe. It is worth knowing what changes outside it, and what does not.

The finding that outranks every group difference

In a study of 585 people running an identical 12-week program, the change in muscle cross-sectional area ranged from −2% to +59%. Strength changes ranged even more widely.

The spread between individuals inside one protocol is larger than the average difference between men and women, or between young and old. Group averages tell you where to start. They do not tell you what you will get.

Sex

Relative hypertrophy is the same. The meta-analysis on this question found no significant difference between men and women following the same protocol (effect size 0.07, p = 0.31, with zero heterogeneity across studies). A more recent Bayesian meta-analysis reached the same conclusion: absolute gains slightly favour men, relative gains are comparable.

Men gain more muscle in kilograms because they start with more muscle and more testosterone. As a percentage of what someone has, the training response is not meaningfully different. The practical implication is that women do not need a different kind of program, and the widespread advice to train lighter with higher reps has no basis in the hypertrophy evidence.

On training around the menstrual cycle: systematic reviews to date do not support programming by cycle phase. The studies are small, methods for confirming phase are inconsistent, and results conflict. Symptoms are real and worth accommodating; a structured phase-based periodization scheme is not currently supported by the evidence, whatever its popularity.

After menopause, the loss of estrogen accelerates decline in both muscle and bone, which raises rather than lowers the value of resistance training. Bone is the outcome with the clearest case, and mechanical loading is one of the few interventions that acts on it directly.

Age

Older adults hypertrophy. The response is blunted rather than absent, and the mechanism has a name: anabolic resistance. Ageing muscle responds less to a given dose of protein and a given dose of loading, so both doses need to be larger.

The most common error in training older adults is under-dosing on the assumption that hard training is unsafe. Resistance training is the primary intervention for sarcopenia, and the risk of doing too little is considerably better documented than the risk of doing too much.

Training age

The single largest predictor of how fast someone gains is how long they have been training. Untrained beginners gain quickly and would gain on almost any program, which is also why beginner results are poor evidence that a given method is superior.

What is still genuinely contested?

Where the volume ceiling sits, the exact weight for indirect sets, training at long muscle lengths, and how much individual variation matters. Stated plainly, because most content in this space presents settled answers where there are none.

What do the drugs and supplements actually do?

“Does it build muscle” and “should you take it” are different questions. Most writing in this space collapses them, usually by implying that anything risky or unapproved must also be ineffective. That is not what the evidence says, and treating a reader as though it does costs you their trust on everything else.

What follows separates the two. Efficacy first, from the trials. Cost second, from the same literature.

Large, well-demonstrated effects

Modest, real effects

Effect on the wrong outcome

Weak, unreplicated, or absent

What the effective ones cost

Anabolic steroids work, and that is precisely why the risk conversation matters. Documented consequences of supraphysiologic androgen use include suppression of endogenous testosterone production lasting months to over a year after cessation, impaired fertility, adverse changes in lipids, elevated hematocrit, and cardiomyopathy with prolonged high-dose use. None of that is speculative, and none of it is an argument that they do not build muscle. It is the trade being made.

SARMs are not the clean alternative they are sold as. The most informative detail in the LGD-4033 trial is not the lean mass gain. It is that the lowest studied dose, 1 mg daily, dropped total testosterone from roughly 663 to 209 ng/dL in three weeks. They suppress the axis substantially, at doses far below what is used recreationally, over durations far shorter. Add documented liver injury, marked HDL suppression, and the fact that chemical analysis of 44 products sold online found only about half contained the labelled compound, with undisclosed anabolic steroids present in some.

So the honest summary on SARMs: modest proven efficacy at doses nobody uses, meaningful axis suppression at those same low doses, unknown effects at real-world doses and durations, and a supply chain that frequently is not selling what it claims.

How to read a supplement claim

Three questions separate most of the signal from most of the noise. Has it been replicated by a group with no stake in the result? HMB is the cautionary example. Was the outcome the one you care about? Growth hormone raises lean mass measures and does nothing for strength. Was the comparison fair? Protein supplementation looks impressive against a low-protein control and marginal against an adequate diet.

The one worth taking

Creatine monohydrate, 3–5 g/day, no loading phase required. It is the only supplement in this space with replicated efficacy across independent groups, a well-characterised mechanism, decades of safety data, and a cost of a few cents a day. It does not raise testosterone and it is not dramatic. It works on its own terms, which is more than the rest of the category can say.

On regulation

Approval status is worth knowing and is a poor proxy for whether something works. Plenty of effective compounds are unapproved for the use in question, and plenty of legal supplements do nothing. Where regulatory facts genuinely bear on a decision they are stated here as what they are: information about legal exposure and product quality, not about biology.

Sources

General educational information about resistance training, not medical advice, and no substitute for assessment by a qualified professional who knows your history. Reading this page does not create a clinician–patient relationship. The auditor performs arithmetic on values you enter using a stated fractional-counting convention; it does not measure anything about you and does not diagnose any condition. Training volume bands describe averages across published studies and individual response varies widely. Content reflects published research current at the time of writing.