The potential influence of postmenopausal estrogen therapy on the subsequent development of non-insulin-dependent diabetes mellitus (NIDDM) is relatively unexplored, despite postulated effects of these hormones on glucose tolerance. We examined the association between postmenopausal hormone use and the subsequent incidence of NIDDM in a prospective cohort of 21,028 postmenopausal US women aged 30 to 55 years and free of diagnosed diabetes, cardiovascular disease, and cancer in 1976. During 12 years of follow-up (422,991 person-years), we confirmed 1249 cases of NIDDM. Current users of postmenopausal hormones had a relative risk of NIDDM of 0.80 (95% confidence interval, 0.67 to 0.96) as compared with never users, after adjustment for age and body mass index (BMI). Past users of these hormones had an age- and BMI-adjusted relative risk of 1.07 (0.93 to 1.23). These results were not materially altered by multivariate adjustment for age, BMI, family history of diabetes, and coronary risk factors. Comparable results were obtained when the analysis was restricted to symptomatic NIDDM as the outcome. We observed no appreciable modification of these associations by family history of diabetes or category of BMI. Duration of current or past use of postmenopausal hormones and dose of conjugated estrogen were not significantly related to incidence of NIDDM. Type of hormone (estrogen alone, progesterone alone, or combination) also did not appear to influence NIDDM risk. These prospective data indicate that postmenopausal hormone therapy is unlikely to be associated with a material increase in the incidence of NIDDM among women.