Our objective was to assess a new quantitative ultrasound device suitable for the measurement of speed of sound in radius. The so-called "bidirectional" technique allows an accurate estimation of velocity based on a compensation for soft tissue effects implemented directly inside the probe. Velocity measurements at 1 MHz of the first arriving signal were performed at the one third distal radius in 358 enrolled women. The average velocity by age decade increases to a peak velocity of 4043 m/s in the class 30-39 y (n = 19) and decreases thereafter. Fracture discrimination was investigated on the subset of the population for which dual-energy x-ray absorptiometry measurement was available, in addition to first arriving signal velocity measurements. The study group consisted of 122 postmenopausal women without history of fracture (group NF) and 44 postmenopausal patients (group F) with osteoporotic fractures (hip, spine, Colles fracture). When adjusted for age and bone mass index, the odds ratio (OR) for fracture prediction by ultrasound velocity, was 1.81 (1.21; 2.70) and OR associated to neck femur BMD was 2.07 (1.31-3.29). For the full model including age and body mass index as cofactors, the area under the receiver operating characteristic curve was 0.77, either for ultrasound velocity or neck femur bone mineral density. Despite the small population and the variety of fractures in the fracture group, our data indicate that the velocity of the first arriving signal measured by bidirectional technique discriminates patients with osteoporotic fracture from controls.