A cohort of 2270 white women, aged 40-69 years at baseline, were followed for an average of 8.5 years in the Lipid Research Clinics Program Follow-up Study. There were 44 deaths due to cardiovascular disease among the 1677 nonusers of estrogens and six cardiovascular disease deaths among the 593 estrogen users. The age-adjusted relative risk (RR) of cardiovascular disease deaths in users compared with nonusers was 0.34 (95% confidence limits 0.12 to 0.81). After multivariable adjustment for potential confounding factors (age, blood pressure, and smoking), the estimated RR for estrogen use was 0.37 (95% confidence limits 0.16 to 0.88). Analyses were done to explore whether these results could be due to selection bias for estrogen use. However, the prevalence of cardiovascular disease at baseline was slightly higher in estrogen users (12%) than in nonusers (10%); furthermore, the exclusion of all women with prevalent cardiovascular disease at baseline did not alter the apparent protective effect of estrogen use on cardiovascular disease mortality (RR = 0.42, 95% confidence limits 0.13 to 1.10). Additional analyses examining the complex association between estrogen use, lipoprotein levels, and cardiovascular disease mortality suggest that the protective effect of estrogen is substantially mediated through increased high-density lipoprotein levels.