We investigated the usefulness of bone density measurements from multiple skeletal sites and calcaneus ultrasound for evaluating the probability of vertebral deformation. Bone mineral density (BMD) was measured at the second metacarpal and middle phalanges using radiographic absorptiometry of hand radiographs, and at the lumbar spine using dual-energy x-ray absorptiometry. Distal radius and proximal radius were measured using single-energy x-ray absorptiometry (SXA), expressed as bone mineral content (BMC, grams per centimeter), and as BMD (grams per square centimeter). The calcaneus was measured using both SXA (BMD) and broadband ultrasound attenuation (BUA). Among the women in this study (mean age 74, SD = 5), 84 women developed new vertebral deformations (57 cases with one and 27 cases with two or more deformations), which were identified on serial radiographs during an average of 9 years prior to the measurements of bone density. Logistic regression analysis was used to calculate odds ratios for risk of deformation corresponding to a 1-SD difference in density or ultrasound, adjusted for age. All bone measurements were significantly associated with vertebral deformation, with odds ratios (95% confidence intervals) ranging from 1.40 (1.10, 1.78) for proximal radius BMD to 1.88 (1.45, 2.44) for calcaneus BMD measurements. Measurements of calcaneal BUA, calcaneal BMD, and hand BMD generally remained significant when included simultaneously with another measurement in the same model, suggesting that spine or radius BMD may not provide much additional information about risk of deformation. It appears that all of the measurements of bone density and ultrasound provide useful information regarding the probability of deformation. These findings await confirmation in a prospective study.