Background: Cycling has been suggested to be related to risk of all-cause and cardiovascular disease (CVD) mortality. However, a quantitative comprehensive assessment of the dose–response association of cycling with risk of all-cause and CVD mortality has not been reported. We performed a meta-analysis of cohort studies assessing the risk of all-cause and CVD mortality with cycling. Methods PubMed and Embase databases were searched for relevant articles published up to December 13, 2019. Randomefects models were used to estimate the summary relative risk (RR) of all-cause and CVD mortality with cycling. Restricted cubic splines were used to evaluate the dose–response association. Results: We included 9 articles (17 studies) with 478,847 participants and 27,860 cases (22,415 from all-cause mortality and 5445 from CVD mortality) in the meta-analysis. Risk of all-cause mortality was reduced 23% with the highest versus lowest cycling level [RR 0.77, 95% confdence interval (CI) 0.67–0.88], and CVD mortality was reduced 24% (RR 0.76, 95% CI 0.65–0.89). We found a linear association between cycling and all-cause mortality (Pnon-linearity=0.208); the risk was reduced by 9% (RR 0.91, 95% CI 0.86–0.96) with each fve metabolic equivalent of task (MET)-h/week increase in cycling. We found an approximately U-shaped association between cycling and CVD mortality (Pnon linearity=0.034), with the lowest risk at approximately 15 MET-h/week of cycling. Conclusions: Our fndings based on quantitative data suggest that any level of cycling is better than none for all-cause mortality. However, for CVD mortality, one must choose an appropriate level of cycling, with an approximate optimum of 15 MET-h/week (equal to 130 min/week at 6.8 MET).