Discontinuation of hormone replacement therapy (HRT) is much more common than what is reported in randomized, double-blind clinical trials. Our purpose in this retrospective study, using a prescription database, was to compare the continuation rate among women who took cyclic combination therapy adding progesterone to estrogen (CYC-PERT) or continuous combined estrogen progestin therapy (CC-PERT). The study subjects were 1,532 women, ≥45 years old, who initially filled index prescriptions for 0.625 mg conjugated estrogens. They were divided into two groups (CYC-PERT = 644, CC-PERT = 888) on the basis of coprescribed medroxyprogesterone. We found that for all women initiating therapy, 35-40% did not return for a refill and 76-81% stopped therapy within 3 years. Those prescribed CC-PERT initially were more likely to stop than those prescribed CYC-PERT (rate ratio [RR] = 1.20; 95% confidence interval [CI] = 1.06-1.35). Adjustments for age, year of starting medication, cost of medication, and prescriber specialty did not affect the difference in discontinuation between the two regimens (RR 1.18, 95% CI = 1.04-1.34). We conclude that the likelihood of women continuing HRT beyond 3 years of initiation is low. Furthermore, compared with CYC-PERT users, those receiving CC-PERT have a slightly higher probability of discontinuation. Efforts should be made to understand why three quarters of women beginning HRT will stop it long before it can provide major long-term benefit.