Exercise may lift students' sleep quality, with caveats

Poor sleep is endemic among university students and is associated with adverse academic, psychological and cardiometabolic outcomes. Exercise is an obvious candidate fix: cheap, scalable and free of the side effects that come with sleep medication. But while previous reviews have examined physical activity and sleep in students generally, evidence from randomised controlled trials of structured exercise programmes measured with the Pittsburgh Sleep Quality Index had not been adequately synthesised. This meta-analysis set out to do exactly that, for both overall sleep quality and its component parts.

The authors searched seven databases from inception to April 2026 for randomised controlled trials comparing structured exercise interventions with non-exercise controls in university students, using PSQI-based outcomes. Seven trials involving 701 participants contributed eight comparisons; six trials with 602 participants entered the primary analysis of the PSQI global score. Mean differences were pooled with random-effects models, and the result was stress-tested with leave-one-out analyses, change-from-baseline analyses, prediction intervals and alternative estimation methods, with certainty graded using GRADE.

Exercise reduced the PSQI global score by 2.43 points versus controls, and the result survived every sensitivity analysis applied. Component analyses in three trials found significant improvements in the sleep-duration and habitual sleep-efficiency scores, while effects on sleep latency and sleep disturbances favoured exercise but were not statistically significant. However, heterogeneity was very high (I² = 91.7%) and the 95% prediction interval ranged from -6.98 to 2.12, crossing the null — meaning the true effect could plausibly be zero in some settings. GRADE certainty was very low for most outcomes and low for sleep efficiency.

The direction of effect matches the broader literature suggesting exercise benefits sleep, and the consistency across multiple sensitivity analyses is encouraging. But the combination of very high heterogeneity, a prediction interval that includes no effect, and very low certainty of evidence means this cannot yet be treated as settled science. The authors call for larger, rigorously designed trials using standardised exercise protocols and objective as well as subjective sleep measures before definitive practice recommendations can be made.

For students sleeping badly, this is a reason to think structured exercise could plausibly help, particularly with how long and how efficiently you sleep. It is not proof: the underlying trials varied enormously and the pooled effect may not translate to every campus or every type of programme. The honest reading is a promising signal from a small evidence base that needs bigger and better trials to confirm.