OBJECTIVE: To conduct an umbrella review of prospective meta-analyses and perform a causal relationship analysis to evaluate causal effects. METHODS: PubMed, Web of Science, Embase, and manual reference list searches were used from database inception to July 27, 2023. Meta-analyses of prospective studies on non-genetic risk factors for breast cancer incidence were included. Overlapping articles were assessed using corrected coverage area. We utilized the AMSTAR-2 criteria to evaluate methodological quality and graded each meta-analysis to assess the strength of evidence. This study is registered with PROSPERO (CRD42023470151). We further explored the causal impacts. RESULTS: Risk factors were classified into 11 categories. Among the 281 meta-analyses of cohort studies, five (1.8%) provided strong evidence, eight (2.8%) indicated highly suggestive evidence, and 23 (8.2%) and 55 (19.6%) showed suggestive and weak evidence, respectively. Breast density (2.89
2.57-3.25), cardiac glycoside (1.39
1.33-1.45), atrial fibrillation (1.18
1.14-1.22), vegetable-fruit-soybean dietary pattern (0.87
0.83-0.92), and postmenopausal women with BMI ≥25 (0.86
0.81-0.91) were strongly associated with breast cancer incidence. For all associations graded as weak evidence or higher, further confirmed the causal relationship between BMI, fruit intake, calcium channel blockers, cheese intake, insulin like growth factor-1 levels, serum triglyceride levels causally. DISCUSSION: Identifying primary risk factors is crucial for delineating high-risk populations among women, facilitating tailored prevention strategies and advancing investigations into underlying mechanisms. SYSTEMATIC REVIEW REGISTRATION: https://www.crd.york.ac.uk/PROSPERO/, identifier CRD42023470151.