Objective To systematically appraise and summarise meta-analyses investigating the effect of exercise compared with a control condition on health outcomes in cancer survivors.

Design Umbrella review of intervention systematic reviews.

Data sources Web of Science, Scopus, Cochrane Library, CINAHL and MEDLINE databases were searched using a predefined search strategy.

Eligibility criteria Eligible meta-analyses compared health outcomes between cancer survivors participating in an exercise intervention and a control condition. Health outcomes were cardiovascular fitness, muscle strength, health-related quality of life, cancer-related fatigue and depression. Pooled effect estimates from each meta-analysis were quantified using standardised mean differences and considered trivial (<0.20), small (0.20–0.49), moderate (0.50–0.79) and large (≥0.80). Findings were summarised using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system.

Results There were 65 eligible articles that reported a total of 140 independent meta-analyses. 139/140 meta-analyses suggested a beneficial effect of exercise. The beneficial effect was statistically significant in 104 (75%) meta-analyses. Most effect sizes were moderate for cardiovascular fitness and muscle strength and small for cancer-related fatigue, health-related quality of life and depression. The quality of evidence was variable according to the GRADE scale, with most studies rated low or moderate quality. Median incidence of exercise-related adverse events was 3.5%.

Conclusion Exercise likely has an important role in helping to manage physical function, mental health, general well-being and quality of life in people undergoing and recovering from cancer and side effects of treatment.

PROSPERO registration number CRD42015020194.