PURPOSE: Several mechanistic studies support a role of cholesterol or its metabolites in breast cancer etiology, but associations have been inconsistent in epidemiological studies. In observational studies, possible reverse causation must be accounted for using a prospective design. We investigated prospective associations between pre-diagnostic serum lipid concentrations [total cholesterol (TC), high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, triglycerides], and both breast cancer risk and survival in the E3N cohort study. METHODS: Analyses were performed on 583 cases from the E3N prospective cohort diagnosed between 1994 and 2005, and 1,043 controls matched on date, age, recruitment center and menopausal status at blood collection. Odds ratios (OR) and 95% confidence intervals were estimated using conditional logistic regression. Risks of recurrence were estimated among cases using Cox proportional hazards model. Models were adjusted for lifestyle risk factors and mutually adjusted for lipid concentrations. Survival analyses were additionally adjusted for tumor characteristics. RESULTS: Overall, there was no association between any serum lipid and breast cancer risk or survival. In stratified analyses, statistically significant interaction was observed between TC and menopausal status (P interaction = 0.05) and between TC and waist circumference (P interaction = 0.03), although the ORs did not reach statistical significance in any of the strata. There was no statistically significant effect modification by BMI, time between blood donation and diagnosis or ER status. CONCLUSIONS: Our results suggest that serum lipids are not associated with breast cancer risk overall, but that menopausal status and waist circumference should be considered in further studies.