Fat intake was examined in relation to risk of breast cancer using dietary questionnaires administered both before and after the diagnosis of breast cancer. Among the Nurses' Health Study cohort members who completed a food frequency questionnaire and were free of cancer in 1986, 398 were diagnosed with breast cancer during 2 years of follow-up. These cases and 798 age-matched controls (cohort members who also completed a dietary questionnaire in 1986 but did not develop breast cancer) were sent another food frequency questionnaire in 1989 inquiring about their diet in 1985. Three hundred cases and 602 controls responded to the second questionnaire. The age-adjusted analysis using the prospective (1986) questionnaire demonstrated no appreciable associations between breast cancer incidence and intakes of total fat (odds ratio (OR) between highest and lowest quintiles = 0.87; 95% confidence interval (CI) 0.54-1.40) and saturated fat (OR = 0.97; CI 0.64-1.46). The age-adjusted analysis using the retrospective (1989) questionnaire suggested positive associations between breast cancer incidence and intakes of total fat (OR = 1.43; 95% CI 0.90-2.27) and saturated fat (OR = 1.38; 95% CI 0.89-2.13). When nutrients were adjusted for total energy intake, the direction and magnitude of the differences between the prospective and retrospective analyses were similar to the analyses unadjusted for energy intake. These results suggest that case-control studies of diet and breast cancer may yield biased associations between fat intake and the risk of breast cancer.