Objectives: To review current publications to examine safety of tai chi (TC). Design: Cochrane Library, EBSCO host and MEDLINE/PubMed were searched for randomized controlled trials (RCTs) including TC as the core intervention and reporting adverse events (AEs). Data were extracted considering active vs. inactive control group comparisons and presence of an AE monitoring protocol. Meta-analyses were conducted for overall results as well as control group and reporting specific conditions. Results: In 256 RCTs of TC, 24 met eligibility criteria (1794 participants) and were assessed using the Cochrane Risk of Bias tool. The frequency of non-serious, serious and intervention-related AEs were not found to be significantly different between TC and inactive or active control conditions. In studies with an AE monitoring protocol, more non-serious adverse events (RD = 0.05; 95% CI: 0.00, 0.10; P = 0.05) were reported for TC compared to inactive interventions. Given the higher overall AE risks related to studies of participants with heart failure, additional analyses examined this set separately. More serious AEs were found for inactive interventions compared with TC in studies with heart failure participants (RD = -0.11; 95% CI: -0.20, -0.03; P = 0.01). Conclusion: Findings indicate that TC does not result in more AEs than active and inactive control conditions, and produces fewer AEs than inactive control conditions for heart failure patients.