Objective: To examine the association between postmenopausal hormone replacement therapy (HRT) and the trajectory of global cognitive change with age. Methods: The Modified Mini-Mental State Examination (MMSE) was administered to a population sample of 2,073 nondemented, community-dwelling female residents of Cache County, UT, aged 65 and older. Current and past HRT and other medications at a baseline interview and at follow-up 3 years later were assessed. Between interviews, a telephone Women’s Health Questionnaire was administered to assess initial exposure, duration, and recency of HRT. Generalized estimating equation marginal models were used to evaluate the cross-sectional and longitudinal relations of HRT and modified MMSE score. Also assessed were effects with multivitamins and calcium supplements as exposures likely to reflect a "healthy lifestyle" among HRT users. Model covariates included the presence of APOE {epsilon}4 alleles, age, education, concurrent depression, several chronic diseases, and self-perceived general health. Results: Age, lower education, depression, and APOE {epsilon}4 were all associated with lower baseline modified MMSE scores. With these covariates in the model, lifetime HRT use was associated with better baseline modified MMSE scores and a slower rate of decline. Stratification by APOE genotype did not alter these effects. Apparent benefits with HRT were attenuated but remained significant after elimination of scores from participants with incident dementia. A significant interaction between age and HRT indicated the strongest effects in women aged 85 and older. Measures of age at initial use of HRT, duration, and recency of exposure did not improve the models. No effects were seen with the "healthy lifestyle" control exposures. Conclusions: In a population cohort of older women, lifetime HRT exposure was associated with improved global cognition and attenuated decline over a 3-year interval. Improvements were greatest in the oldest old.