Postmenopausal women often discontinue estrogen therapy within the first year. No studies have examined reasons why women continue or discontinue hormone replacement therapy (HRT) after several years of use, when hormone side effects have subsided. We wanted to identify determinants of estrogen use after participation in a 3-year randomized placebo-controlled trial. The Postmenopausal Estrogen/Progestin Intervention (PEPI) study compared the effects of estrogen alone or in combination with one of three progestogens with placebo. Post-PEPI use of hormone therapy was ascertained 1-4 years after the trial in 775 (90%) of the originally enrolled women. Potential correlates of treatment decisions were identified from sociodemographic characteristics, medical histories, and clinical measures ascertained at baseline and during and after the trial. Among women who had been assigned to placebo during PEPI, post-PEPI hormone use was significantly less common in women who were adherent to placebo during PEPI, older, or of non-Caucasian ethnicity. Hormone use was positively associated with hysterectomy. Among women assigned to an active regimen during PEPI, post-PEPI hormone use was significantly more common in women who used hormones before PEPI and in women who were adherent to hormones during PEPI. Older age, less education, and being non-Caucasian predicted less hormone use. Post-PEPI hormone use was highest in San Diego and lowest in Iowa City. Women on placebo who lost more bone mineral density (BMD) were more likely to begin hormones than women with less bone loss. Lipids, blood pressure, and other cardiovascular risk factors had relatively little influence on hormone use. The main predictors of post-PEPI hormone use were those associated with use in the general population (education, ethnicity, geographical region, hysterectomy, and prior use/adherence).