Abstract
BACKGROUND: Cognitive impairment is common among individuals with depression and contributes to functional impairment and poor treatment outcomes. Exercise, as a non-pharmacological intervention, has been increasingly recognized as a promising approach for improving cognitive function. However, the dose-response relationship underlying these effects remains poorly understood, and the mechanisms through which exercise may influence cognition or depressive symptoms have yet to be clarified. To address these gaps, this study used a three-level meta-analysis to quantify the effects of exercise on cognitive function in depression, examine potential moderators and dose-response patterns, and conduct an exploratory mediation analysis testing whether cognitive changes mediate exercise-related improvements in depressive symptoms. METHODS: We searched China National Knowledge Infrastructure (CNKI), Wanfang Data, Web of Science, Embase, the Cochrane Library, and PubMed for experimental studies evaluating the effects of exercise interventions on cognitive function in patients with depression, from database inception to May 30, 2025. Risk of bias was assessed using the Cochrane Risk of Bias 2 (RoB 2) tool, and certainty of evidence was graded with GRADEpro. Three-level meta-analyses were conducted with the metafor package in R. RESULTS: Thirty-one studies, including 2,324 participants with depression, were included. Exercise type included aerobic, resistance/strength training, mind-body practices, and multicomponent exercises. Exercise frequency ranged from 1 to 5 sessions per week; intervention duration from 3 to 16 weeks; and exercise intensity was categorized as low, moderate, or moderate to high. Session duration ranged from 30 to 180 min. A three-level meta-analysis showed that exercise improved cognitive function (Hedges’ g (g) = 0.24; 95% confidence Interval (CI), 0.14 to 0.33; p < 0.001). Moderator analyses indicated significant influences of exercise intensity, inpatient status, intervention content, study design, and weekly exercise time (all p < 0.05). Moderate-intensity interventions produced the largest gains (g = 0.36; 95% CI, 0.23 to 0.50; p < 0.05). Dose-response modelling suggested that weekly exercise volumes below approximately 176 min may yield limited benefit (prediction interval crossing zero). In an exploratory mediation analysis, the indirect effect of exercise on depressive symptoms via cognition was insignificant (a×b = 0.02; 95% CI, -0.00 to 0.04), corresponding to a proportion mediated of 5.3%. CONCLUSIONS: Exercise is associated with improvements in cognitive function among individuals with depression. Rather than indicating a universal recommended dose, the dose–response analysis revealed a minimum weekly activity threshold, with cognitive benefits becoming more likely when moderate-intensity exercise exceeds approximately 176 min per week. Current evidence does not support cognitive improvement as the primary mediating pathway through which exercise influences depressive symptoms, and this finding should be interpreted cautiously. Further multicenter randomized controlled trials with rigorous methodology and standardized cognitive assessments are needed to refine modality-specific dose-response patterns and clarify potential mediating pathways. TRIAL REGISTRATION: PROSPERO registration number CRD42025645011. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12966-026-01902-3.