Background: Prior studies have reported an inverse association between physical activity (PA) and risk of heart failure (HF). However, a comprehensive assessment of the quantitative dose-response association between PA and HF risk has not been reported previously. Methods and results: Prospective cohort studies with participants >18 years of age that reported association of baseline PA levels and incident HF were included. Categorical dose-response relationships between PA and HF risk were assessed with random-effects models. Generalized least-squares regression models were used to assess the quantitative relationship between PA (metabolic equivalent [MET]-min/wk) and HF risk across studies reporting quantitative PA estimates. Twelve prospective cohort studies with 20 203 HF events among 370 460 participants (53.5% women; median follow-up, 13 years) were included. The highest levels of PA were associated with significantly reduced risk of HF (pooled hazard ratio for highest versus lowest PA, 0.70; 95% confidence interval, 0.67-0.73). Compared with participants reporting no leisure-time PA, those who engaged in guideline-recommended minimum levels of PA (500 MET-min/wk; 2008 US federal guidelines) had modest reductions in HF risk (pooled hazard ratio, 0.90; 95% confidence interval, 0.87-0.92). In contrast, a substantial risk reduction was observed among individuals who engaged in PA at twice (hazard ratio for 1000 MET-min/wk, 0.81; 95% confidence interval, 0.77-0.86) and 4 times (hazard ratio for 2000 MET-min/wk, 0.65; 95% confidence interval, 0.58-0.73) the minimum guideline-recommended levels. Conclusions: There is an inverse dose-response relationship between PA and HF risk. Doses of PA in excess of the guideline-recommended minimum PA levels may be required for more substantial reductions in HF risk. Keywords: exercise; heart failure; meta-analysis; prevention and control.