AimsWe aimed to determine the effects of different exercise modalities in patients with type 2 diabetes mellitus (T2DM). MethodsWe searched PubMed, Embase, and the Cochrane Library from their inception until July 2020 to identify randomised controlled trials (RCTs) on exercise in adults with T2DM. Paired reviewers independently performed study selection, data extraction, and risk of bias assessment. The certainty of the evidence was assessed using the Confidence in Network Meta-Analysis (CINeMA) framework. ResultsA total of 106 RCTs that enroled eight exercise modalities with 7438 patients were included. Six exercise modalities, except unsupervised aerobic/resistance exercise, significantly reduced glycosylated haemoglobin (HbA1c), with mean differences (MDs) ranging from 0.71 (95% confidence interval [CI]: 0.34-1.08) to 0.34 (95% CI: 0.17-0.52), low to high certainty, in comparison with no exercise. The evidence of low to moderate certainty showed that supervised aerobic/resistance exercise improved glycaemic control, body weight, blood pressure, and blood lipid profiles compared with no exercise. Flexibility exercise may be associated with glycaemic control (HbA1c: MD = 0.71, 95% CI: 0.34-1.08); fasting plasma glucose (MD = 1.48, 95% CI: 0.78-2.17), and weight loss (MD = 1.80, 95% CI: 0.85-2.75) compared with controls, but not blood pressure and lipid profiles. Balance exercise showed the largest benefit in improving total cholesterol (MD = 52.81, 95% CI: 28.47-77.16) and low certainty. We found no significant differences between exercises and the triacylglycerol (TG) level. ConclusionsOverall, our network meta-analyses support the recommendation for exercise in patients with T2DM, especially supervised exercises. Limited evidence supports the benefits of flexibility and balance exercises. The effectiveness of exercise modalities for TG reduction remains unclear.