Aims: Whether Tai Chi (TC) benefits patients with type 2 diabetes mellitus (T2DM) remains controversial. Thus, we performed a meta-analysis to assess the efficacy of TC in T2DM patients. Methods: A computerised search through PubMed and Embase was performed to identify relevant studies. The primary outcomes were fasting blood glucose (FBG), haemoglobin A1c (HbA1c) and insulin resistance (HOMA). Secondary outcomes included total cholesterol, high-density lipoprotein cholesterol (HDL-C) and triglyceride. Weighted mean differences (WMDs) and 95% confidence intervals (CIs) were calculated. Results: 4 randomised controlled trials (RCTs) and 5 non-randomised controlled trials (NRCTs) met the inclusion criteria. The pooled WMDs from RCTs were -14.82 mg/dL (95% CI: -49.17 to 19.53; P=0.40) for FBG, -0.19% (95% CI: -0.41 to 0.03; P=0.09) for HbA1c and -0.34 units (95% CI: -3.02 to 2.34; P=0.80) for HOMA. The WMDs from NRCTs were -11.22 mg/dL (95% CI: -18.58 to -3.86; P=0.003) for FBG, -0.41% (95% CI: -0.53 to -0.29; P<0.00001) for HbA1c and -0.60 units (95% CI: -1.46 to 0.25; P=0.16) for HOMA. Furthermore, the pooled results of serum lipids suggest that TC significantly reduced triglyceride (P=0.006) instead of total cholesterol (P=0.77), and failed to improve HDL-C (P=0.12). Conclusions: Sufficient evidence to support the benefits of TC to T2DM patients is lacking. Further large-scale studies are needed to investigate the long-term efficacy of TC.