OBJECTIVE: To determine the relative risk for sustaining a first hip fracture after hormone replacement therapy. DESIGN: Prospective population-based cohort study with an average observation period of 5.7 years. SETTING: A prescription-based cohort in the Uppsala health care region in Sweden. PARTICIPANTS: The cohort (23 246 women) comprised virtually all women of 35 years of age and older who received noncontraceptive estrogens from April 1977 through March 1980. Comparisons were made with women in the background population. MEASUREMENTS: Follow-up through 1983 was done with regard to hospital admissions for a first cervical or trochanteric hip fracture. The observed number of cases was compared with that expected on the basis of person-years of observation in the cohort and incidence rates in the background population. Analyses were made in exposure categories, based solely on prescription data. MAIN RESULTS: During 133 022 person-years of observation, 163 cases of first hip fracture occurred, compared with the 205.5 expected, yielding an overall relative risk of 0.79 (95% CI, 0.68 to 0.93). The greatest protective effect (relative risk, 0.37; CI, 0.13 to 0.79) was found against trochanteric fracture among women receiving potent estrogens who were under 60 years of age at cohort entry. This group also had the highest proportion of treatments with combinations of estrogens and progestogen (41%). Treatment with less potent estrogens, mainly estriols, had no protective effect. Data indicated that the baseline risk for hip fracture was not lower in the cohort women than in the background population. CONCLUSIONS: Treatment with potent estrogens, both alone and possibly when combined with progestogens, reduces the risk for both cervical and trochanteric hip fractures within the first decade after menopause.