Aim To analyse the effects of active video games on physical function in independent community-dwelling older adults. Design Systematic review and meta-analysis of randomized controlled trials. Data sources The CINAHL, LILACS, Medline, Proquest and Scopus databases were consulted, with no restriction by year of publication. Review methods Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed. The meta-analysis was performed using RevMan software. Results The analysis included 22 randomized controlled trials with a total of 1208 participants (all >= 55 years old). In our meta-analyses, the effects produced by playing the active video games (mean differences) were statistically significant for the variables Gait speed and Timed up-and-go. The differences between the control and experimental groups were not significant in the following tests: 6-minute walk, 30-second chair stand, balance (measured with the Berg Balance Scale), cadence, grip strength, knee extension strength, 8-Foot Up-and-Go or velocity. Conclusions Physical exercise from participation in active video games has beneficial effects on two clinical parameters (Gait speed and Timed up-and-go) in independent community-dwelling older adults. However, the effects on other parameters do not differ from those obtained with conventional exercise training. Therefore, the clinical significance of these benefits is limited. Impact Older adults usually perform little physical activity. In consequence, researchers have increasingly considered alternatives to traditional forms of exercise. One such is that provided by active video games, which can be a source of stimulation, encouraging adherence and motivation in exercise programmes. Our review shows that active video games can improve gait speed and mobility, but in other respects obtain no differences from conventional exercises. Further tailored randomized clinical trials should be undertaken with diverse populations of older adults to evaluate different physical function variables to determine the most appropriate training approach and its optimal design and duration.