Tell athletes a dummy pill is caffeine and many get faster and stronger. A review of 17 studies puts a number on the effect

Caffeine is one of the few supplements with solid evidence behind it for sport. But part of what it does may come from expectation rather than chemistry. A systematic review and meta-analysis published online on 20 October in the Journal of the International Society of Sports Nutrition tried to measure that part on its own, by gathering studies in which people were deceived: they were given a dummy capsule, saline, decaffeinated coffee or gum and told it contained caffeine.

The team, from universities in the UK, Brazil and the Netherlands, searched four databases up to July 2026 and screened 1,292 records. Seventeen studies met their rules, which included a baseline or no-treatment comparison so that a placebo effect could actually be estimated. Together they covered 257 healthy adults with a median age of 23. The tests ranged from 1,000-metre and 4,000-metre running time trials and 10 km and 40 km cycling time trials to bench presses, leg extensions to failure and a 60-metre sprint. Most studies used capsules and timed them about an hour before the test, when real caffeine would be peaking.

Pooled together, believing they had taken caffeine improved performance with a standardised effect size (Cohen's d) of 0.49, with a 95% confidence interval of 0.28 to 0.69 — a small-to-moderate effect. The authors note that this sits inside, and at times above, the range of 0.16 to 0.63 reported for real caffeine in an earlier umbrella review of 21 meta-analyses. In secondary analyses, the largest pooled effect was on time to exhaustion (d 1.07, but from only two studies and 17 people). Perceived effort, repetitions and total weight lifted also improved, while power output, velocity and heart rate showed no clear change.

Two findings stand out. Effects were larger when the researchers running the test did not know the capsule was a placebo, which the authors suggest may be because staff who know can unconsciously signal low expectations. And studies with more women showed smaller effects, though only five of the 17 studies included any women. The review's own certainty rating was low: nine studies were judged at high risk of bias and eight unclear, and no study was rated low risk. The practical message is modest — what athletes expect from a supplement can shape how they perform, and that expectation is part of what any caffeine trial measures.