Commonly known risk factors have been shown to correlate with mean bone mass in populations, but none of these risk factors, either alone or in combination, have been shown to be predictive of future fracture risk in an individual. In order to evaluate the predictive power of bone mineral measurements at various sites, bone mineral content (BMC) has been measured at four skeletal sites and compared to subsequent fracture incidence at all skeletal locations, including spine and appendicular sites. Os calcis BMC has the most consistent monotonic relationship to unadjusted fracture incidence rates; os calcis BMC also has the strongest (p = 0.009) relation to levels of relative risk, after adjustment for age, height, and weight. Based on actual ability to predict fracture risk prospectively, along with such secondary criteria as cost, ease of performance, precision, and radiation exposure, the os calcis appears to be an optimal BMC measurement site for routine screening of perimenopausal women.