Several meta-analyses examining the effects of exercise on depression have been criticized for including studies of poor methodological integrity. More recent meta-analyses addressed the most common criticism by including only randomized control trials; however, these analyses suffer from incomplete literature searches and lack of moderating variable analyses. Using a more extensive search procedure, the current meta-analysis examines the effects of exercise on depressive symptoms in 58 randomized trials (n = 2982). An overall effect size of −0.80 indicates participants in the exercise treatment had significantly lower depression scores than those receiving the control treatment. This frac34; SD advantage represents level 1, Grade A evidence for the effects of exercise upon depression. Analysis of moderating variables examined the influence of population characteristics, exercise characteristics and methodological characteristics. Examination of clinical significance in 16 trials with clinically depressed patients found 9 of 16 exercise treatment groups were classified as ‘recovered’ at post-treatment, with another three groups classified as ‘improved’. Analysis showed dropout rates for the exercise treatment were similar to those found in psychotherapeutic and drug interventions.