Background Depression and anxiety are non-motor symptoms of Parkinson’s disease (PD). Physical exercise is a promising approach to reducing neuropsychological burden. We aimed to comprehensively synthesize evidence regarding the use of exercise for treating depression and anxiety symptoms in PD. Methods Systematic review and meta-analysis following PRISMA recommendations. Searches on PubMed, Cochrane Library, Scopus, Web of Science, Embase, and Physiotherapy Evidence Database (PEDro) was conducted. The random-effects model was employed for all analyses with the standardized mean difference as the effect estimate. Results Fifty records were retrieved, but only 17 studies met the criteria for the meta-analyses. A moderate to large effect was observed for depression (−.71 [95% CI = −.96 to −.46], 11 studies, 728 individuals), and a small to moderate effect for anxiety (−.39 [95% CI = −.65 to −.14], 6 studies, 241 individuals), when comparing exercise to non-exercise controls. Subgroup analysis revealed significant effects from aerobic (−.95 [95% CI = −1.60, −.31]), mind-body (−1.85 [95% CI = −2.63, −1.07]), and resistance modalities (−1.61 [95% CI = −2.40, −.83]) for depression, and from mind-body (−.67 [95% CI = −1.19 to −.15]) and resistance exercises (−1.00 [95% CI = −1.70 to −.30]) for anxiety. Conclusion Physical exercise has a relevant clinical impact on depression and anxiety in PD. We discuss the level of the evidence, the methodological limitations of the studies, and give recommendations.