Dance-based interventions have been proposed for the management of Parkinson's disease (PD) symptoms. This review critically appraises and synthesises the research on the effects of dance interventions on gait, cognition and dual-tasking in PD, through a meta-analysis of peer-reviewed literature from seven databases. Eligible studies included people with PD, used a parallel-group or cohort design with a dance-based intervention, reported outcome measures of gait, cognition or dual-tasking, and were published in English up until September 2017. Of the initial 1079 articles, 677 articles were reviewed for eligibility, and 25 articles were retained. Only 12 articles had sufficient common assessment items for meta-analysis. Two independent reviewers extracted the data and assessed the risk of bias of each study using the Cochrane risk-of-bias tool. Based on pre-post change scores, gait speed, Timed Up and Go (TUG) test performance, freezing of gait questionnaire, and six-minute walk test times significantly improved after a dance intervention compared to controls. Global cognition assessed with Montreal Cognitive Assessment, and cognitive dual-tasking measured using dual-task TUG, also exhibited greater improvement in dance groups. There was limited evidence to determine the most effective intensity, frequency, duration of dance interventions or the most beneficial music. Findings must be interpreted cautiously because of the lack of randomised control trials, and the moderate to high risk of bias of studies. However, the results of papers with level-I and level-II.1 evidence suggest that dance may have the potential to ameliorate PD symptoms, particularly gait, global cognition and cognitive dual-tasking.