The objective of the study was to determine the incidence rate of osteoporotic fractures among elderly women who had long-term postmenopausal estrogen replacement therapy (ERT) and to compare this with the incidence rate in women who had not used estrogen. In a previous retrospective cohort study based on medical record review in 1982, we showed that long-term ERT was associated with lower incidence of wrist and vertebral fractures. We have extended our follow-up of 490 women by adding a mean 8 years to the observation period, which more than triples the number of osteoporotic fractures. At the Kaiser Permanente Medical Center, San Francisco, a large health maintenance organization, a review of computer pharmacy records from 1968 through 1971 identified 245 postmenopausal women; all had begun estrogen within 3 years of menopause and had used estrogen for at least 5 years. From the same pharmacy records, 245 age-matched postmenopausal non-users were identified. Among estrogen users, mean length of use was 17.0 years, mean follow-up after treatment was 7.3 years and mean dose of conjugated oral estrogen was 0.9 mg daily. We found statistically significant reduction in the incidence of wrist and vertebral fractures in users compared with non-users. The age-adjusted incidence ratios (95% confidence intervals for wrist fractures were 0.55 (0.32-0.92) and for vertebral fractures were 0.57 (0.41-0.80). These results were not statistically significantly altered after adjustment for age of menopause, body mass index and smoking. It is concluded that long-term ERT confers statistically significant protection against wrist and vertebral fractures.