Postmenopausal estrogens can delay or prevent osteoporosis and decrease the frequency of fractures, but they also increase the risk of endometrial cancer. A decision analytic model was developed using a Markov process with 18 different states to estimate quality-adjusted life expectancy with and without estrogen therapy. The model considered fractures of the hip, wrist, pelvis, humerus and spine with potential outcomes of short-term morbidity, long-term disability, nursing home placement, and death. Quality adjustments were based on expert opinions. In sensitivity analyses, various risks of endometrial cancer and hysterectomy due to estrogens were examined. The effect of estrogen therapy on cardiac mortality also was considered. For a cohort of 50-year-old white women who would take estrogens for 15 years, the analysis showed a benefit of 0.47 years but 0.67 quality-adjusted years. For every five-fold increase in the relative risk of endometrial cancer, the benefit decreases by 0.07 quality years. The benefit would increase by 0.17 quality years for each 10 percent decrease in the fracture rate and 0.32 for each 10 percent decrease in cardiovascular mortality rate. Thus, estrogen therapy provides a significant gain in quality-adjusted life expectancy. In considering the efficacy of any drug, all the benefits of the drug as well as all its risks must be included. If the beneficial effect of estrogens on cardiovascular mortality is confirmed, it will overshadow all other effects. Any recommendation about postmenopausal estrogens with respect to osteoporosis that excludes their cardiovascular effects markedly underestimates the potential gains from therapy.