Research summary · Nutrition and supplements
Caffeine showed a pooled swimming benefit, but evidence certainty remained low
This review included thirteen randomized crossover studies involving 144 men and 48 women. A three-level model accounted for multiple performance measurements within studies rather than treating all 28 effect sizes as independent experiments.
- Publication year
- 2026
- First published online
- Publication type
- Systematic review and three-level meta-analysis of placebo-controlled crossover trials
- Journal
- Journal of the International Society of Sports Nutrition
- Population / context
- Swimming · Competitive swimmers · Women and men
What the publication reports
The pooled performance effect favoured caffeine (standardized mean difference 0.57, 95% CI 0.20–0.94). An alternative analysis translated this to about a 1.7% improvement. Effects varied substantially between studies. Exploratory subgroup results suggested a larger effect at higher doses, but they do not establish the best dose for an individual swimmer.
Limits and interpretation
The authors rated certainty low, and every included trial had some risk-of-bias concerns. There were relatively few women, small studies and varied swimming protocols. Subgroup patterns are hypothesis-generating; the review does not justify escalating caffeine intake or ignoring sleep, anxiety and other adverse effects. A pooled laboratory benefit is not a guaranteed race improvement.
Read the source
Consult the original record for the full author list, methods, references, funding and competing-interest disclosures.
Caffeine makes a splash: a systematic review and multilevel meta-analysis exploring the effects of caffeine intake on swimming performance
Original publication / indexed recordPMID: 42323844DOI: 10.1080/15502783.2026.2692016Full text at the sourceReader discussion
Discuss the evidence.
Ask a question, share a source or discuss the methods. Respectful disagreement is welcome. Every comment and reply is reviewed before publication.
Loading comments…