Training Volume and Frequency: What the New Dose-Response Analysis Found

More weekly resistance-training volume was associated with greater muscle and strength gains in a new dose-response analysis, but the benefit flattened as volume rose. Frequency followed a different curve: its modeled relationship with hypertrophy was compatible with a negligible effect, whereas strength showed a positive relationship with diminishing returns.

That direct answer comes from a Sports Medicine systematic review and meta-regression, not a trial that assigned one large sample to every possible dose. Its curves describe average study-level relationships. They do not output an ideal weekly set count for an individual.

Two curves, two outcomes

Volume and frequency often move together in real programs, yet they are not the same variable. Weekly volume describes how much work is performed. Frequency describes how often that work is distributed. Adding a third weekly session could raise total work, simply spread the same work across more days, or do some of both.

The researchers assembled 67 studies containing 2,058 participants. The pooled sample was 79.1% male, and the average age was 25.16 years. Multilevel models for hypertrophy and strength adjusted for study duration and training status.

It does not create the control of a single trial where people are randomized to precisely defined doses. Exercise selection, effort, loading, supervision, measurement methods, and the populations studied still differed among the underlying papers. The result is a map of the literature, not a personal dose finder.

For volume, the posterior probability that the marginal slope exceeded zero was 100% for both outcomes. More volume aligned with larger gains, but each addition was expected to return less than the one before it. The flattening was considerably stronger for strength than hypertrophy. “Diminishing returns” does not identify a universal ceiling; it describes the changing slope of the modeled relationship.

Frequency separated the outcomes. For hypertrophy, the positive-slope probability was below 100%, leaving the estimate compatible with a negligible effect. For strength it was 100%, again with diminishing returns. More frequent practice of a tested lift could matter for strength independently of muscle growth, while frequency can also make a fixed workload easier to distribute. The model cannot completely separate those roles.

The accounting problem hidden inside a set total

The paper's most interesting methodological choice was how it handled exercises that train a muscle directly versus indirectly. A biceps curl is direct work for the elbow flexors. A row may train them indirectly while directly targeting other muscles. Treating both as identical sets can exaggerate the apparent weekly dose to the biceps; ignoring the row can discard real work.

The team tested three accounting methods:

• Total counting: an indirect set counted as 1.
• Fractional counting: an indirect set counted as 0.5.
• Direct counting: an indirect set counted as 0.

Relative evidence was strongest for fractional counting, so the primary models used that method. This does not prove every compound-exercise set is exactly half a set for every assisting muscle. It shows that, across the included studies, the middle assumption described the data better than always counting indirect work fully or not at all.

For program review, the idea is more valuable than the decimal. A weekly set total only means something when the exercises behind it are clear. Twelve sets of direct quad work and 12 sets in which the quads contribute to different degrees are not necessarily equivalent exposures.

The fractional method fit best, so it drove the primary models. It is an audit convention, not proof that every compound set supplies exactly half a set to every assisting muscle. The practical lesson is to inspect the exercises behind a weekly number before comparing programs. The detailed model is available in the published paper.

A worked interpretation, not a prescription

Imagine a lifter audits weekly biceps work: six rows and three pull-ups train the biceps indirectly, while three curls train them directly. Total counting labels all 12 as biceps sets; direct-only counting credits only the three curls. For a fractional estimate, counting each of the nine compound sets as 0.5 yields 4.5 direct-equivalent sets; adding the three direct curl sets gives 7.5.

The 0.5 coefficient is a bookkeeping assumption used for this audit, not an evidence-derived universal conversion for every exercise or lifter. The 7.5 total is therefore a transparent program description, not a biological measurement. If progress stalls, the lifter can identify whether a proposed addition is direct work, indirect exposure, or merely another session containing the same work.

Suppose those sets currently occur on one day. Splitting them across two days changes frequency without necessarily raising volume. That may improve session quality or provide another strength-practice exposure, but this analysis does not prove the split itself adds hypertrophy. Adding two direct sets changes volume and perhaps frequency together. The progressive-overload study provides separate context for changing training inputs over time, while the ACSM guidance offers broader programming context.

The sensible use of the curves is incremental. Describe current work, make a small change for a defined reason, and judge the response rather than chasing the highest modeled dose. A lifter already progressing with productive sessions has no result here requiring more work.

Where the model stops answering

Meta-regression inherits the differences among its source studies: exercise choice, loading, effort, supervision, measurement, and populations varied. The evidence was mostly young and male. Frequency often redistributed volume, which complicates causal interpretation, and study-level associations cannot reveal one person's response curve.

The authors reported no conflicts specific to this article but disclosed that all five also work as fitness-industry coaches and writers. That context should be reported without treating it as evidence of misconduct.

Most importantly, the paper supplies no hard threshold at which useful work becomes wasteful. Its diminishing curves warn that the next set is unlikely to be as valuable as the first few; they do not tell a reader where to stop. That decision remains outside what this model can establish.