Investigations of the effects of estrogen replacement on cognitive function in healthy older women have yielded disparate results. We evaluated the relationship between a history of estrogen use and cognitive test performance in 727 women participating in a large community-based study. Participants were followed longitudinally for an average of 2.5 years. Estrogen use history was obtained at baseline. Standardized tests of memory, language, and abstract reasoning were administered at baseline and at follow-up. Results indicate that women who had used estrogen replacement scored significantly higher on cognitive testing at baseline than nonusers, and their performance on verbal memory improved slightly over time. The effect of estrogen on cognition was independent of age, education, ethnicity, and APOE genotype. Results suggest that estrogen replacement therapy may help to maintain cognitive function in nondemented postmenopausal women.