Background: Parkinson’s disease (PD) is a neurodegenerative disorder that presents with motor and nonmotor symptoms such as bradykinesia, resting tremor, postural instability, and cognitive and neuropsychiatric manifestations. Dance therapy or complex motor activity, besides pharmacological treatment, may have benefits in PD patients. Objective: To assess the effect of dance in patients with PD. Methods: We searched for clinical trials in PubMed, Scopus, and Web of Science, and Cochrane till April 2020 using relevant keywords. Data were extracted and pooled as mean difference (MD) with 95% confidence interval (CI) by Review Manager 5.3. Results: Fourteen randomized controlled trials with 372 patients were included. Dance showed a significant improvement over the control group in term of the Unified Parkinson’s Disease Rating Scale III (UPDRS III) after three (MD = − 4.49, 95% CI [− 6.78, − 2.21], p = 0.00001), six, (MD = − 5.96, 95% CI [− 8.89, − 3.02], p < 0.0001), and 12 months (MD = − 14.58, 95% CI [− 24.76, − 4.4], p = 0.005), and Mini-BES test after 12 months. Compared to exercise, dance showed a significant improvement in Timed Up and Go (TUG) test, Berg Balance Scale (BBS), and Mini-BES test. Conclusion: In comparison to other types of exercise or no activity, dance improves the symptoms and outcomes in patients with PD, especially motor symptoms. Dance also has positive effects on balance, functional mobility, and cognition.