Objective: There is a high prevalence, yet under-treatment of depressive disorder and symptoms by conventional therapy in people with multiple sclerosis (MS). We conducted a meta-analysis examining the overall effect of exercise training on depressive symptoms in MS. Methods: We searched PubMed for randomized controlled trials (RCT) of exercise training and depression as an outcome in samples with MS. There were 13 RCTs that met inclusion criteria and yielded data for effect size (ES) generation (Cohen's d). An overall ES was calculated using a random effects model and expressed as Hedge's g. Results: The weighted mean ES was small, but statistically significant (Hedge's g=0.36, SE=0.09, 95% CI=0.18-0.54, z=3.92, p<.001) indicating the exercise training resulted in an improvement in depressive symptoms compared to control. The overall effect was not heterogeneous (Q=16.46, df=12, p=0.17, I2=27.08); and post-hoc, exploratory analyses only identified depression symptom scale as a potential moderator variable (p=0.04). Conclusion: The cumulative evidence indicates that exercise training can yield a small, yet statistically significant and reliable reduction in depressive symptoms for people with MS. © 2014 Elsevier Inc.