As part of a prospective study begun in 1981, we evaluated 8,600 postmenopausal women and 5,049 men residing in a southern California retirement community for risk factors for hip fracture. Incidence rates were twice as high in women as in men, but in both sexes the rates nearly doubled every 5 years between 70 and 90 years. Active exercise was strongly and negatively associated with hip fracture risk in both sexes; the age-adjusted relative risk was 0.6 and 0.5 for females and males, respectively, for 1 or more hours of exercise per day compared with less than 1/2 hour of exercise. A high body mass index (upper tertile of weight divided by height squared) was associated with a strong reduction in hip fracture risk for females (RR = 0.5). Current cigarette smokers had a significantly increased risk (RR = 1.8 and RR = 2.2 for females and males, respectively) compared with never-smokers, but the risk for past smokers was not different from that of lifetime nonsmokers. Other factors related to reduced hip fracture risk in women were high parity, late age at menarche, and long menstrual cycle length. These age-adjusted relative risk estimates did not change materially in multivariate analysis when adjusted simultaneously for age, active exercise, body mass, smoking, and, for women, age at menarche and number of children. Among estrogen users, the lowest risk of hip fracture was observed for recent users (RR = 0.8), while users who had stopped estrogen use 15 or more years ago had a relative risk of 1.1, suggesting that the protective effect of estrogen dissipates after many years since cessation of estrogen therapy.