Sets per session, or sets per week? A weekly-adjusted clarification of the per-session hypertrophy dose–response

How many sets should be performed in a single training session matters for programming and recovery, so the per-session dose–response analysis of Remmert et al. (2025) has understandably attracted attention. Because the two papers discussed here come from the same five authors — Remmert, Pelland, Robinson, Hinson and Zourdos, in different orders — we offer this comment in the spirit of a clarification within one research programme, not a dispute between camps. We also begin with a correction of our own. Remmert et al. report a positive per-session dose–response for hypertrophy with diminishing returns, and locate the point of undetectable outcome superiority (PUOS) at approximately 11 fractional sets per session. They are explicit that the PUOS is not a ceiling or upper limit: data at very high per-session volumes are sparse, and whether still higher volumes attenuate, or further improve, adaptation is unknown. A figure of around 16 sets per session (range 13–24) that has circulated in discussion of their work was not theirs; it came from our own earlier, unpublished refit, and we attribute it to ourselves and withdraw any contrary impression. The question that remains is narrower and, we think, constructive: does the per-session association survive once weekly volume is held fixed? Remmert et al.'s caution that the PUOS should not be read as an optimum already anticipates part of the answer. Their open data — for which we thank them — allow the rest to be checked directly. The difficulty is arithmetic before it is statistical. In the group's shared arm-level dataset, per-session dose is weekly sets divided by training frequency, so the two doses cannot vary independently: arms with many sets per session are, in general, arms with many sets per week (in these data the correlation between log per-session dose and log weekly volume is 0.83). An unadjusted per-session curve therefore inherits part of the weekly dose–response that Pelland et al. (2026) establish. The informative model is the joint one, in which weekly volume is held fixed and the per-session coefficient asks whether the packaging of a given weekly volume — more sets in fewer sessions, versus fewer sets in more — is itself associated with hypertrophy. Because weekly volume is per-session dose multiplied by frequency, that coefficient is equivalently the frequency coefficient at fixed weekly volume, with sign reversed.