The ability of quantitative ultrasound (QUS) to estimate the risk of osteoporotic fractures was evaluated in a prospective study over a mean time of 5.47 years in 254 postmenopausal women (mean age 58.06+/-7.67 years). Baseline measurements of ultrasound transmission velocity (UTV) and bone mineral density (BMD) were taken at the distal radius (DR). UTV was also measured at the patella (P). Fifty nonspine fractures due to minor trauma were detected during annual check-ups with an incidence of 3.59/year. Fractures occurred in older women with a lower BMD and QUS. Using Cox regression analysis the relative risk (RR) per 1 standard deviation (SD) decrease in the unadjusted QUS and BMD measurements was: BMD-DR = 3.56, 95% confidence interval (CI) 1.57-8.09; 95% CI 2.08-9.68. The relationship between BMD and QUS variables and fracture risk persisted after adjusting for potential confounders apart from previous fractures, giving the following RR: BMD-DR = 2.99, 95% CI 1.06-8.41; UTV-DR = 3.69, 95% CI 1.18-11.49; UTV-P = 3.89, 95% CI 1.53-9.90. Correcting also for previous fractures, only UTV-P remained an effective predictor of fracture risk even after QUS measurement correction for BMD. Wrist fractures were best related to BMD-DR (RR 7.33, 95% CI 1.43-37.50) and UTV-DR (RR 10.94, 95% CI 1.10-108.45), while hip and ankle fractures were significantly associated only with UTV-P (hip: RR 32.14, 95% CI 1.83-562.80; ankle: RR 17.60, 95% CI 1.78-173.79). The combined use of BMD and QUS is a better predictor of fracture risk than either technique used separately. Comparison of the areas under the receiver operating characteristic (ROC) curves did not show differences in the ability of BMD and QUS to correctly distinguish fractures. In conclusion, QUS predicts fracture risk in osteoporotic women at least as well as BMD. UTV-DR and BMD-DR are good predictors of wrist fractures, while UTV-P is strongly related to hip and ankle fractures. QUS and BMD combined improve the diagnostic ability of each technique individually.