Objective: To assess the effectiveness of exercise in adults with clinical depression. Data sources: The databases CINAHL, Embase, Cochrane Database of Systematic reviews, Cochrane Controlled Trials Register, MEDLINE, and PsycINFO were searched (1806-2008) using medical subject headings (MeSH) and text word terms depression, depressive disorder and exercise, aerobic, non-aerobic, physical activity, physical fitness, walk*, jog*, run*, bicycling, swim*, strength, and resistance. Study selection: Randomized trials including adults with clinical depression according to any diagnostic system were included. Data extraction: Two investigators evaluated trials using a prepiloted structured form. Data synthesis: Thirteen trials were identified that fulfilled the inclusion criteria. Eight had adequate allocation concealment, 6 had a blinded outcome, and 5 used intention-to-treat analyses. The pooled standardized mean difference (SMD) calculated using a random-effects model was -0.40 (95% CI, -0.66 to -0.14), with evidence of heterogeneity between trials (I(2) = 57.2%, P = .005). There was an inverse association between duration of intervention and the magnitude of the association of exercise with depression (P = .002). No other characteristics were related to between-study heterogeneity. Pooled analysis of 5 trials with long-term follow-up (ie, that examined outcomes beyond the end of the intervention) suggested no long-term benefit (SMD, -0.01; 95% CI, -0.28 to 0.26), with no strong evidence of heterogeneity in this pooled analysis (I(2) = 23.4%, P = .27). There was no strong statistical evidence for small study bias (P > .27). Only 3 studies were assessed as high quality (adequately concealed random allocation, blinded outcome assessment, and intention-to-treat analysis). When we pooled results from these, the estimated beneficial effect of exercise was more modest (SMD, -0.19; 95% CI, -0.70 to 0.31) than the pooled result for all 13 studies, with no strong evidence of benefit. Conclusions: Our results suggest a short-term effect of exercise on depression: on average, depression scores 0.4 of a standard deviation lower in clinically depressed patients randomly assigned to an exercise intervention at the end of that intervention compared to those randomly assigned to a none exercise group. There is little evidence of a long-term beneficial effect of exercise in patients with clinical depression.