Studies of humans have not confirmed the suggestion from animal studies that estrogen replacement therapy may have an inverse relation with cognitive function decline. Because many of these studies have been marred by design or methodological problems, such as a small sample size, failure to control for confounding variables, or the use of a cross-sectional design, the present study was conducted in a large cohort of middle-aged postmenopausal women participating in the Atherosclerosis Risk in Communities (ARIC) Study. The study population consisted of 2,859 women aged 48-67 years, whose cognitive function was tested at the second (1990-1992) and fourth (1996-1998) visits of the ARIC Study using three instruments: the Delayed Word Recall Test, Digit Symbol Subtest of the Wechsler Adult Intelligence Scale-Revised, and Word Fluency Test. After multiple adjustment, no consistent patterns of cognitive changes between the two cohort visits could be detected according to current use or duration of use of estrogen replacement therapy. Thus, the results of the present study do not support the hypothesis that estrogen replacement therapy may slow age-related cognitive decline, at least as it applies to relatively young postmenopausal women.