The association between menopausal estrogen therapy and hip fracture was studied in a retirement community. Ninety-one hip fracture cases during a 5-year period in female residents under age 80 were compared to age-and race-matched community controls. Estrogen use was recorded from the medical records of the outpatient care facility and personal interviews. The estimated risk ratio for use of oral estrogens in excess of 60 months was 0.42. This protective effect was largely limited to oophorectomized women for whom the risk ratio for a comparable duration of use was 0.14; the risk significantly decreased with increased duration, but no such trend existed with increased dosage. Diabetes mellitus, low Quetelet's index, tallness, prolonged immobilization or physical inactivity, use of corticosteroids, early age at menopause, low levels of sunlight exposure, and heavy cigarette smoking were each independent risk factors for hip fracture but none confounded the observed association with estrogen use.