PURPOSE Postdiagnosis exercise is associated with lower breast cancer (BC) mortality but its link with risk of recurrence is less clear. We investigated the impact and dose-response relationship of exercise and recurrence in patients with primary BC. METHODS Multicenter prospective cohort analysis among 10,359 patients with primary BC from 26 centers in France between 2012 and 2018 enrolled in the CANcer TOxicities study, with follow-up through October 2021. Exercise exposure was assessed using the Global Physical Activity Questionnaire-16, quantified in standardized metabolic equivalent of task–hours per week (MET-h/wk). We examined the dose/exposure response of pretreatment exercise on distant recurrence-free interval (DRFI) for all patients and stratified by clinical subtype and menopausal status using inverse probability treatment weighted multivariable Cox models to estimate hazard ratios (HRs). RESULTS For the overall cohort, the relationship between exercise and DRFI was nonlinear: increasing exercise ≥ 5 MET-h/wk was associated with an inverse linear reduction in DRFI events up to approximately 25 MET-h/wk; increasing exercise over this threshold did not provide any additional DRFI benefit. Compared with <5 MET-h/wk, the adjusted HR for DRFI was 0.82 (95% CI, 0.61 to 1.00) for ≥ 5 MET-h/wk. Stratification by subtype revealed the hormone receptor–/human epidermal growth factor receptor 2– (HR–/HER2–; HR, 0.59 [95% CI, 0.38 to 0.92]) and HR–/HER2+ (HR, 0.37 [95% CI, 0.14 to 0.96]) subtypes were preferentially responsive to exercise. The benefit of exercise was observed especially in the premenopausal population. CONCLUSION Postdiagnosis/pretreatment exercise is associated with lower risk of DRFI events in a nonlinear fashion in primary BC; exercise has different impact on DRFI as a function of subtype and menopausal status.