Measurement of ultrasonographic parameters provides information concerning not only bone density but also bone architecture. We investigated the usefulness of ultrasonographic parameters and bone mineral density for evaluating the probability of vertebral fracture. 397 postmenopausal women (59.1 +/- 6.0 years) with (n = 178) or without (n = 219) atraumatic vertebral fractures were studied. In all women, bone mineral density (BMD) of the lumbar spine was evaluated by dual X-ray absorptiometry (DXA) and speed of sound (SOS); broadband ultrasound attenuation (BUA) and Stiffness in the calcaneus were evaluated by an Achilles unit (Lunar Corporation). Ultrasonographic parameters and BMD were compared by examining the magnitude of the odds ratios, to determine which produces the highest estimate of the probability of odds of fracture, and by examining widths of the respective confidence intervals (CI) to show which estimate of odd ratio is the most precise. The relative risk of vertebral fracture, after adjusting for potential confounders, was 3.5 (CI 2.6-4.8) for BUA; 4.5 (CI 3.2-6.2) for SOS; 5.8 (CI 4.0-8.4) for Stiffness and 7.5 (CI 4.8-11.5) for BMD. Ultrasound (US) parameters were still significant independent predictors of vertebral fracture, even after adjusting for BMD. The relative risk of fracture for a simultaneous decrease by 1 SD of BMD and by 1 SD of each ultrasound parameter was 17.3 (CI 9.4-39.6) for BMD and SOS; 18.3 (CI 8.4-30.6) for BMD and BUA and 22.1 (CI 8.9-52.7) for BMD and Stiffness. Our data suggest that US and BMD provide complementary information which can be combined to improve estimates of vertebral fracture risk.