Loading fundamentals 03: Training frequency

Short answer. With weekly volume held constant, training a muscle more often produced no more growth or strength. On our reading, frequency matters as the way volume gets delivered: once your weekly sets exceed what one session can productively hold, spread them across sessions, which for most people means two or three exposures per muscle a week.

---

What it is

How many times per week you train a given muscle. The debate has run for decades between the once-a-week bodypart split and higher-frequency arrangements.

Frequency is easy to isolate on paper: hold weekly sets constant, change only the distribution. That design has been run enough times to pool.

The proposed mechanism

The case for higher frequency rests on muscle protein synthesis. In young men performing twelve sets of elbow flexion, synthesis rose 50% at four hours, reached 109% above baseline at twenty-four, and returned to within 14% of the untrained arm by thirty-six (MacDougall et al., 1995). A systematic review put the ceiling around 48 hours.

If the elevation lasts a day or two, training once a week leaves most of the week at baseline, so higher frequency means more activations of mTOR, the signaling hub that switches on muscle protein synthesis, for the same total work.

Two findings complicate it. Repeated bouts appear to desensitize the signal: in a rodent model, mTORC1 activity six hours after a second bout was significantly lower than after the first. And in mice given bouts separated by 72, 24 or 8 hours, protein synthesis was activated in the 72- and 24-hour conditions but not the 8-hour one.

Both that account and the competing one, that a concentrated session allows more work with fuller recovery between, predict a volume-equated difference.

What has actually been tested

Hypertrophy, volume equated. A meta-analysis pooled 25 studies and 836 participants and found no significant difference between higher and lower frequency. The same held for direct measures of growth and for resistance-trained participants (Schoenfeld et al., 2019).

Hypertrophy, volume not equated. In the same analysis, studies that did not equate volume showed a significant effect favoring higher frequencies, modest between one and three-plus days per week.

Let higher frequency bring more volume with it and it wins. Hold volume constant and it does not.

Strength. Same pattern, separate analysis. Across 22 studies, effect sizes (in standard deviations) rose from 0.74 to 0.82 to 0.93 to 1.08 across one, two, three and four-plus sessions per week. That looks like a clean dose-response. The subgroup restricted to volume-equated studies found no significant effect at all (Grgic et al., 2018). An ascending series that disappears when you control the confounder is instructive, because the ascending series is what gets quoted.

The specific claim that two beats one. It existed, and from the same lead author. An earlier meta-analysis pooled 10 studies and concluded that on a volume-equated basis, twice a week produced superior hypertrophy to once, and that major muscle groups should be trained at least twice weekly (Schoenfeld et al., 2016).

The 2019 analysis, same team, 25 studies instead of 10, did not reproduce it. Same question, more data, opposite conclusion, three years apart. The "at least twice a week" rule now repeated almost everywhere is the 2016 conclusion, still traveling after its own authors revised it.

Volume itself does have a dose-response, with slope estimates in the region of a quarter to four tenths of a percent additional hypertrophy per additional weekly set. Volume and frequency have distinct dose-response relationships, and how indirect sets are counted materially changes the answer (Pelland et al., 2026).

And a single session appears to have a ceiling. A meta-regression of 67 studies and 2,058 participants examined per-session set volume. The posterior probability, the analysis's estimated chance given the data, that the slope exceeded zero was 100% for both outcomes, so more sets in a session does produce more. But both best-fitting models showed diminishing returns, with the point of undetectable outcome superiority at around eleven fractional sets per session (indirect sets counted as half) for hypertrophy and two direct sets for strength (Remmert et al., 2025). This is a preprint, not peer reviewed.

The two findings interact, and that is the point. The volume-equated frequency trials mostly ran at low to moderate per-session volumes. If returns flatten above ten sets, those trials were all operating below the ceiling, where distribution should not matter. That may be the only region tested.

Our reading

Frequency is not a growth variable in its own right. It is the mechanism by which weekly volume becomes deliverable, and it stops being optional once your weekly volume exceeds what one session can productively hold.

The mechanistic argument is not the one to lean on. More frequent mTOR activation is a reasonable prediction from the time-course data and was the original rationale for the higher-frequency movement. But it predicts a volume-equated difference, and that difference is not there. With the desensitization findings added, the mechanism arrives where most mechanisms in this publication arrive: real at the cellular level, not doing the work people credit it with.

The per-session ceiling is the argument that survives. If returns within a session flatten around ten or eleven fractional sets, the weekly volume you can deliver is bounded by frequency multiplied by that ceiling. Someone targeting twenty hard sets per week cannot fit them into one session and have them all count, because the later sets are performed by a muscle that has already flattened out, on top of the capacity loss documented elsewhere in this branch.

This reframes the null result rather than contradicting it. Frequency does not matter when your weekly volume fits comfortably in one session, and that is the condition most of the research was conducted under.

It also partly rescues the 2016 conclusion by a different route. Training a muscle once a week forces every weekly set into one session, which is the arrangement most likely to run past the ceiling. Twice may well beat once. Not because two exposures are physiologically better than one, but because one exposure cannot carry the volume.

So the practical position is this. Set weekly volume first, then pick the frequency that delivers it without any session running far past the point where sets stop counting. At low weekly volumes that is a free choice and the meta-analyses say so. At high weekly volumes it is not a choice, and the once-a-week bodypart split becomes a structural problem rather than a preference. For most people that lands at two or three exposures per muscle per week.

The ceiling argument combines two analyses not designed to be combined, one a preprint. Treat it as an interpretation.

Frequency is the third lever on one quantity: how much non-degraded work you accumulate in a week. (→ Loading fundamentals 01: Rest between sets; Loading fundamentals 02: Reps in reserve; Loading fundamentals sub-hub) The reason to distribute volume across exposures is not the protein synthesis window. It is how many of the week's sets are performed by a fresh muscle.

Where the goals diverge

Limited training days. The volume-equated evidence is reassuring: two full-body sessions carry a respectable weekly volume at no cost. What matters more is how much of each session still counts. (→ Loading fundamentals 01: Rest between sets)

High weekly volume. (→ Loading fundamentals 07: Weekly volume) Here the choice disappears. Twenty sets in one session is not the same twenty spread over three, and distribution also decides how many exercises fit per muscle. (→ Loading fundamentals 06: Exercise selection) It is the region the frequency literature has least to say about, because few trials went there.

Adherence. The authors landed on personal preference as the deciding factor for a given volume, which concedes that the physiology has run out of things to say.

The case against

The frequency literature is largely one author network. Both meta-analyses, the superseded earlier version, and the volume dose-response work share overlapping authorship. When one group produces the finding, the reversal and the reanalysis, the self-correction is real and the external check is missing.

Volume equating is artificial. Nobody equates volume in practice. If higher frequency reliably produces more volume in the real world, the uncontrolled comparison may be the more externally valid one, and we treated it as confounded.

The trained-participant data is thin, mostly untrained or recreational lifters over six to thirty weeks.

Our central argument rests on a preprint. The ceiling figures have not been peer reviewed. If that analysis moves, our argument moves with it, and we are left with the null result and no story about when it stops applying.

The ceiling number is not one to use. Eleven fractional sets is a group-level point of undetectable superiority, not a personal limit.

What is not known

Whether frequency matters at very high weekly volumes, where the ceiling starts binding. Our reading depends on this and it is the least tested question here.

Where any individual's ceiling sits, and whether the protein synthesis rationale measures something on a timescale these studies did not look at.