Long-term postmenopausal hormone therapy alleviates menopausal symptoms, decreases women's risk of osteoporosis and has been shown to reduce cardiovascular morbidity and mortality in more than 30 observational studies. However, nearly half of all women initiating hormone therapy discontinue within the first year. This study was conducted to determine the rate and predictors of hormone therapy discontinuation in a clinical practice setting. We identified 992 women aged 45-59 who began hormone therapy between 1993 and 1995 in a Massachusetts health maintenance organization. Women were followed 2 years from the day they filled a prescription for estrogen. 53% discontinued hormone therapy by the end of the first year and one-fifth stopped after the first prescription. A prescription from a gynecologist (RR = 0.82, 95% CI: 0.68, 0.99) and a mammogram a year prior to initiation (RR = 0.75, 95% CI: 0.63, 0.89) were associated with a lower rate of discontinuation. Women who were using monoamine oxidase inhibitors (MAOI) or selective serotonin reuptake inhibitors (SSRI) antidepressants 3 months prior to initiation (RR = 2.07, 95% CI: 1.26, 3.39) or who had been enrolled in the health plan for less than 3 years (RR = 1.33, 95% CI: 1.10, 1.62) had an increased risk of discontinuing hormone therapy. The year a woman entered the cohort was also associated with a higher rate of discontinuation (RR = 1.40, 95% CI: 1.14, 1.74 for 1994 and RR = 1.95, 95% CI: 1.52, 2.50 for 1995). The results indicate that long-term hormone therapy use is uncommon in clinical practice, particularly when a woman or her physician attempts to use hormone therapy as an alternative to antidepressant therapy, and that the rates of discontinuation of hormone therapy were rising rapidly in the mid-1990s.