To assess the prospective relationship between body mass and rate of hip fracture, we analyzed data from 2,285 postmenopausal women, aged 50-77 years, who responded to the first National Health and Nutrition Examination Survey (NHANES I) in 1971-1974 and the 1982-84 interview of the National Health Epidemiologic Follow-up Study (NHEFS). We plan to add to the previous analysis of these same data by Farmer et al. by 1) including retrospective information on estrogen replacement therapy; 2) increasing the follow-up time by 5 years and using only hospital-verified cases of hip fracture from the 1987 Health Care Facilities Stay data tape; and 3) limiting the analysis to postmenopausal white women, ages 50-77 years. Results of the multivariable Cox regression modeling showed that women with a baseline body mass index in the highest quartile (> 37 kg/m1.5) experienced a 70% lower rate of hip fracture compared with women in the lowest quartile (< or = 28.7 kg/m1.5)(RR=0.32; 95%CI:0.12, 0.82). Age was positively related to the risk of hip fracture, with the rate over 20% higher per year of age (RR=1.21; 95%CI:1.13, 1.29). Although reported education level, smoking history, physical activity level, and estrogen replacement were significantly (p < 0.0001) associated with body mass index, these covariates were not related to hip fracture in the multivariable analysis. Our findings corroborate earlier results using this same data source, and suggest that interventions aimed at preserving lean mass and consequent bone integrity should be encouraged among women before menopause and maintained through older adulthood.