All cases of endometrial cancer occurring among the residents of an affluent retirement community were compared with controls chosen from a roster of all women in the same community. Evidence of estrogen and other drug use and of selected medical conditions was obtained from three sources: medical records of the principal care facility, interviews, and the records of the local pharmacy. The risk ratio for any estrogen use was estimated from all available evidence to be 8.0 (95 per cent confidence interval, 3.5 to 18.1). and the for conjugated estrogen use to be 5.6 (95 per cent confidence interval, 2.8 to 11.1). Increased risk from estrogens was shown for invasive as well as noninvasive cancer, and a dose-response effect was demonstrated. For an estrogen user, the risk from endometrial cancer appeared to exceed by far the base-line risk from any other single cancer.