REF-1626 · Krieger, 2009 · Journal of Strength and Conditioning Research, Vol. · Paper · PubMed

Two to three sets per exercise were associated with 46% greater strength gains than one set in a meta-regression of 14 studies (92 effect sizes; effect-size difference 0.25, CI 0.14-0.37), in trained and untrained subjects alike; 4-6 sets were not clearly better than either 1 set or 2-3 sets

Main caveat: Written from the abstract and the PubMed record; full text not read, so the included studies, their quality and their populations are not known here.
moderate evidenceMeasured on non-athletes to trained athletesRead from the abstract only

Krieger pooled English-language studies that compared one set per exercise with more than one while holding the other training variables equal, to see how the number of sets affects dynamic strength. A hierarchical random-effects meta-regression took in 92 effect sizes, nested in 30 treatment groups from 14 studies.

More than one set gave a larger effect than one set (difference 0.26). Split by dose, 2-3 sets beat 1 set (difference 0.25), while 4-6 sets did not differ significantly from either 1 set or 2-3 sets, and their intervals were wide. The benefit of extra sets did not depend on programme length, on whether subjects were trained, or on upper versus lower body.

No single study drove the result and no sign of publication bias was found. The author's conclusion: 2-3 sets per exercise go with 46% greater strength gains than one (an association, as the abstract words it), in trained and untrained people.

Participants

Number of participants, ages and sex counts: not in the abstract.
Training status: both trained and untrained subjects, per the abstract.
Both sexes by the PubMed indexing (MeSH Male and Female); adults (MeSH Adult, Young Adult).

Protocol

Eligibility: English-language studies comparing single with multiple sets per exercise, with other variables controlled.
Model: hierarchical random-effects meta-regression, effect sizes nested within treatment groups within studies (92 / 30 / 14).
Outcome: dynamic strength.
Dose-response model: 1 set vs 2-3 sets vs 4-6 sets per exercise.
Moderators tested: programme duration, training status, upper vs lower body.
Checks: sensitivity analysis for studies that sway the result; publication bias.
Not in the abstract: databases and search dates, quality appraisal, how many studies fed the 4-6-set estimate, programme lengths, weekly frequency, loads, rest periods and whether sets went to failure.

Key Results

ComparisonDifference in effect size (± as given)CI (level not stated in the abstract)p
Multiple sets vs 1 set0.26 ± 0.050.15 to 0.37< 0.0001
2-3 sets vs 1 set0.25 ± 0.060.14 to 0.370.0001
4-6 sets vs 1 set0.35 ± 0.25−0.05 to 0.740.17
4-6 sets vs 2-3 sets0.09 ± 0.20−0.31 to 0.500.64
2-3 sets per exercise: 46% greater strength gains than 1 set, in trained and untrained subjects (the abstract's summary figure; how the percentage was derived is not in the abstract).
The direction of the 0.09 difference is not stated; the two estimates against one set (0.35 and 0.25) put 4-6 sets slightly ahead (our arithmetic), with an interval that includes zero.
Set number did not interact with programme duration, training status or upper vs lower body.
Sensitivity analysis: no study changed the result on its own; no evidence of publication bias.
Not in the abstract: the effect size of one set on its own, and any strength change in kilograms or percent of 1RM.

Practical Application

Two to three sets beat one. 2-3 sets gave 46% more gain than one (the single-set gain itself is not reported; that one set raised strength is our reading of the percentage) (effect-size difference 0.25, CI 0.14-0.37), in trained and untrained people alike. When time allows, 2-3 sets per exercise is the default this paper supports; when it does not, one set is the fallback (inference; this abstract does not test one set against none, see REF-586 and REF-1141).
Above three sets the data are too thin to tell. 4-6 sets were not significantly better than 1 set (0.35, CI −0.05 to 0.74) or than 2-3 sets (0.09, CI −0.31 to 0.50). Read this as "not shown", neither as "no extra gain" nor as "harmful": the point estimate sits slightly above 2-3 sets, with a wide interval.
Inference, not tested here: for an endurance athlete who lifts, 2-3 sets of each main lift is a sensible target when building, and one set the fallback in a crowded week. The paper measured strength, not cycling or running performance; how the base sets the cyclist's and the runner's gym dose is in SYNTH-43 and SYNTH-44.
The unit is sets per exercise per session. Weekly frequency is not in the abstract, so turning this into weekly set totals needs the weekly-volume and frequency meta-analyses (REF-1623, REF-1624, REF-1477).
Not from this paper; the base's own practice: whatever the set count, the next session's load and reps come from the last sets done and their RPE (SYNTH-101).

Limitations

Written from the abstract and the PubMed record; full text not read, so the included studies, their quality and their populations are not known here.
No risk-of-bias or quality assessment is reported in the abstract; single-author analysis (whether a second person screened or extracted data is not in the abstract).
The 4-6-set estimate is imprecise (± 0.25, against ± 0.06 for 2-3 sets), so it can neither confirm nor exclude a further gain.
Outcome is dynamic strength only; muscle size, power and sport performance were not analysed.
English-language studies only, published up to 2009; the newer weekly-volume meta-analyses in Cross-References pool later trials.
How this entry was checked and changed
Why this level

risk of bias not read (abstract only): a meta-analysis of controlled comparisons, capped at B+ because the abstract reports sensitivity and publication-bias checks but no quality or risk-of-bias assessment of the included studies; the search method is not in the abstract either

Content correction 2026-11-06
2026-11-06: added from the abstract and the PubMed record; level B+ (meta-analysis; risk of bias not read (abstract only)).