The objectives of the study were firstly to determine the accuracy of ultrasound calcaneal measurements in the prediction of bone mineral density determinations with dual-energy X-ray absorptiometry (DXA) of the lumbar spine (LS), femoral neck (FN), and Ward's triangle (FW) in a mixed population of 1000 women, unsorted as to diagnosis, and secondly to determine the accuracy of the various site-specific measurements in predicting each other. Ultrasound measurements [stiffness, speed of sound (SOS) and broadband ultrasound attenuation (BUA)] were made with the Lunar Achilles device, and the bone mineral density (BMD) of the LS, FN and FW were determined with the Lunar DPX. The data were analyzed for correlation, sensitivity, specificity, and accuracy of various paired sites. The coefficients of correlation of the young adult t-scores in the total group between calcaneal stiffness and BMDs of the LS, FN, and FW varied between 0.53 and 0.60. Coefficients for LS versus FN and FW were 0.70 and 0.62, respectively. A comparison of SOS and BUA values obtained at the calcaneus with BMDs of the LS, FN and FW yielded correlation coefficients that varied from 0.54 to 0.56. The general accuracy of prediction of one site by another ranged from 64.2% to 74.4%, where normality was defined as a t-score > -2. It is concluded that no site can predict the status of another site with sufficiently high accuracy to be clinically useful. The role of ultrasound transmission in bone as a predictor of fracture risk is theoretically promising, but has yet to be proved by a long term prospective study.