In a prospective study of 8881 postmenopausal female residents of a retirement community in southern California, we evaluated in detail the relationship between estrogen use and overall mortality. After 7 1/2 years of follow-up, there had been 1447 deaths. Women with a history of estrogen use had 20% lower age-adjusted, all-cause mortality than lifetime nonusers (95% confidence interval, 0.70 to 0.87). Mortality decreased with increasing duration of use and was lower among current users than among women who used estrogens only in the distant past. Current users with more than 15 years of estrogen use had a 40% reduction in their overall mortality. Among oral estrogen users, relative risks of death could not be distinguished by specific dosages of the oral estrogen taken for the longest time. Women who had used estrogen replacement therapy had a reduced mortality from all categories of acute and chronic arteriosclerotic disease and cerebrovascular disease. This group of women had a reduced mortality from cancer, although this reduction was not statistically significant. The mortality from all remaining causes combined was the same in estrogen users and lifetime nonusers.