To avoid biases due to control selection and the retrospective reporting of diet that may occur in case-control studies, a series of large prospective studies have been started since about 1980 to examine the relationship between fat intake and breast cancer risk. Relative risks for the highest versus lowest categories of energy-adjusted fat intake in the six published studies have all been nonsignificant and close to 1.0: California Seventh-Day Adventists Study (215 cases), relative risk (RR) = 1.21; Nurses' Health Study (1439 cases), RR = 0.90; Canadian Breast Cancer Screening Study (592 cases), RR = 1.30; Iowa Women's Study (408 cases), RR = 1.13; New York State Cohort (359 cases), RR = 1.00; and the Dutch Cohort (471 cases), RR = 1.11. Two additional studies have been reported with fewer than 70 cases and dietary methods less appropriate for studies of this nature; an inverse association was seen in one and a positive association in the other. In studies that included both premenopausal and postmenopausal women, no evidence was seen for a stronger positive association among the older group. In addition, no indication was seen of a threshold for reduced risk below 30% of calories from fat, as has been suggested by some. Although associations with specific types of fat tended to vary somewhat among studies, in none was saturated fat significantly associated with risk of breast cancer. In summary, the evidence from prospective studies strongly suggests that a substantial association does not exist between fat intake by middle-aged women and their risk of breast cancer. A very weak association, an effect of very low fat intake, a greatly delayed effect, and an influence of fat consumption early in life are difficult to exclude. Further follow-up of these and other cohorts will provide additional resolution of these issues.