BACKGROUND: Typically, management of PCOS focuses on lifestyle changes (exercise and diet), aiming to alleviate symptoms, and lower the associated risk of type 2 diabetes and cardiovascular disease. Our objective was to analyse evidence on the effectiveness of exercise in the management of PCOS, when compared to (i) usual care, (ii) diet alone, and (iii) exercise combined with diet, and also exercise combined with diet, compared to (i) control or usual care and (ii) diet alone. METHODS: Relevant databases were searched (June 2017) with no time limit for trial inclusion. Eligible trials employed a randomised or quasi-randomised design to measure the chronic effects of exercise, or exercise and diet in women with PCOS. RESULTS: Searches returned 2390 articles; of those, 27 papers from 18 trials were included. Results are presented as mean difference (MD) and 95% confidence intervals (95% CI). Compared with control, exercise had a statistical effect on change from baseline fasting insulin (MD - 2.44 μIU/mL, 95% CIs - 4.24 to - 0.64; very low-quality evidence), HOMA-IR (- 0.57, - 0.99 to - 0.14; very low-quality evidence), total cholesterol (- 5.88 mg/dL, - 9.92 to - 1.83; low-quality evidence), LDL cholesterol (- 7.39 mg/dL, - 9.83 to - 4.95; low-quality evidence), and triglycerides (- 4.78 mg/dL, - 7.52 to - 2.05; low-quality evidence). Exercise also improved VO(2) max (3.84 ml/kg/min, 2.87 to 4.81), waist circumference (- 2.62 cm, - 4.13 to - 1.11), and body fat percentage (- 1.39%, - 2.61 to - 0.18) when compared with usual care. No effect was found for change value systolic/diastolic blood pressure, fasting glucose, HDL cholesterol (all low-quality evidence), or waist-to-hip ratio. Many favourable change score findings were supported by post-intervention value analyses: fasting insulin (- 2.11 μIU/mL, - 3.49 to - 0.73), total cholesterol (- 6.66 mg/dL, - 11.14 to - 2.17), LDL cholesterol (- 6.91 mg/dL, - 12.02 to - 1.80), and VO(2) max (5.01 ml/kg/min, 3.48 to 6.54). Statistically lower BMI (- 1.02 kg/m(2), - 1.81 to - 0.23) and resting heart rate (- 3.26 beats/min - 4.93 to - 1.59) were also revealed in post-intervention analysis. Subgroup analyses revealed the greatest improvements in overweight/obese participants, and more outcomes improved when interventions were supervised, aerobic in nature, or of a shorter duration. Based on limited data, we found no differences for any outcome between the effects of exercise and diet combined, and diet alone. It was not possible to compare exercise vs diet or exercise and diet combined vs diet. CONCLUSION: Statistically beneficial effects of exercise were found for a range of metabolic, anthropometric, and cardiorespiratory fitness-related outcomes. However, caution should be adopted when interpreting these findings since many outcomes present modest effects and wide CIs, and statistical effects in many analyses are sensitive to the addition/removal of individual trials. Future work should focus on rigorously designed, well-reported trials that make comparisons involving both exercise and diet. SYSTEMATIC REVIEW REGISTRATION: This systematic review was prospectively registered on the Prospero International Prospective Register of Systematic Reviews ( CRD42017062576 ).