Reports of a role of postmenopausal estrogen replacement therapy in the development of breast cancer have been inconsistent. Although many epidemiologic studies have failed to show an association between short-term use of estrogen and breast cancer, there are indications that long-term use may present an increased risk. We undertook a long-term, retrospective cohort study of the incidence of breast cancer in women who had taken long-term estrogen (average 17.2 years), compared to women who had not taken estrogen. Subjects were 454 women born between 1900 and 1915, who were members of a large health maintenance organization in northern California. By the end of 1995, 26 (11.2%) of estrogen users developed breast cancer, as did 9 (4.1%) of the nonusers; the relative risk (RR) for estrogen use was 2.8 [95% confidence interval (95% CI) 1.3-5.9]. Adjustment for age and multiple breast cancer risk factors, including breast cancer surveillance, reduced the RR for estrogen to 2.0 (95% CI 0.9-4.5). We conclude that long-term estrogen use is associated with a substantially increased risk of breast cancer.