Physical exercise and esports performance: an updated systematic review and study-level meta-analysis

Physical exercise has been proposed as a means of supporting esports-relevant cognition and gaming performance, but the evidence is heterogeneous and frequently reports multiple correlated outcomes. We updated a systematic review of controlled interventions and reconstructed the meta-analysis with the study as the independent unit. The original searches covered PubMed, Scopus and PsycINFO through May 2026; supplementary searching on 3 August 2026 identified additional controlled evidence, and the database/citation searches were rerun during revision through 4 September 2026, with no further eligible studies identified. Twelve controlled study clusters (660 enrolled or randomized; 608 reported analysed) were included. Nine evaluated exercise-only interventions; five estimable exercise-only studies (184 enrolled; 182 analysed) contributed 18 outcomes to quantitative synthesis after within-study aggregation (assumed between-outcome correlation rho = 0.50), whereas four exercise-only studies were synthesized narratively because compatible sampling variances or effect metrics were unavailable. The primary interpretive focus was placed on outcome domains: cognitive/psychomotor outcomes yielded Hedges' g = 0.465 (95% CI 0.073–0.856; k = 4), whereas gaming-specific outcomes were imprecise (g = 0.574, 95% CI − 0.002 to 1.150; k = 3). An exploratory cross-domain omnibus composite yielded g = 0.512 (95% CI 0.214–0.810; k = 5). Because only five of nine exercise-only studies were statistically convertible, this pooled signal may be affected by availability bias and does not summarize the complete eligible exercise evidence. Four additional exercise trials, two active-virtual-reality trials and one multicomponent trial were synthesized narratively. Evidence certainty was very low. Although I-squared and tau-squared were estimated as zero in the quantitative models, with only k = 3–5 independent studies these heterogeneity statistics have low power and should not be interpreted as evidence that effects are truly homogeneous across esports titles, exercise protocols or outcome constructs. Exercise may benefit selected cognitive or psychomotor outcomes relevant to esports, while evidence for improvement in actual gaming performance remains uncertain; the repeated-exercise subgroup (k = 2) is descriptive only, and no definitive performance prescription or acute-versus-repeated superiority claim is supported.