The effect of dietary calcium on hip fracture risk was examined prospectively using the NHANES I Epidemiologic Follow-Up Study cohort, which is derived from a nationally representative sample of the United States population. A cohort of 4342 white men and postmenopausal women ages 50-74 years at baseline (1971-1975) were observed through 1987 for up to 16 years of follow-up. Quantitative estimates of calcium intake were obtained at baseline from a 24-h recall, while weekly frequency of dairy food consumption was obtained from a qualitative food frequency. By 1987, 44 men and 122 women had experienced a hip fracture according to hospital records or death certificates. In the total sample of women the risk of hip fracture was only slightly lower for the highest quartile compared with the lowest. However, although not statistically significant, the age-adjusted risk of hip fracture was approximately 50% lower in the highest quartile of calcium intake compared with the lowest quartile in the subgroup of women who were at least 6 years postmenopausal and not taking postmenopausal hormone. The low relative risk observed among men, although interesting, must be interpreted cautiously due to small sample size. Adjusting for other risk factors did not appreciably change the results for either sex. The pattern of relative risks for calcium quartiles and by selected cutpoints was not consistent with a dose-response effect of calcium. Our results suggest that calcium may lower hip fracture risk in late menopausal women.