Caucasian postmenopausal women with vertebral fracture (38), hip fracture (38) and without fracture (124) were measured by spine and femoral neck bone mineral density (BMD) (Norland XR 36), and two quantitative ultrasound (US) or QUS parameters (Igea DBM sonic 1200): the amplitude-dependent speed of sound (Ad-Sos) and the US bone profile score (UBPS), at the proximal hand phalanxes to investigate the ability of QUS to detect previous fracture. Age, height and weight were not different among groups, but menopausal age was lower with vertebral fracture (p < 0.005). QUS parameters of hip (AdSos = 1806.2 +/- 104.6 m s(-1); UBPS = 0.22 +/- 0.13) or spine fracture (AdSos = 1774.4 +/- 90.0 m s(-1); UBPS = 0.19 +/- 0.10) were not significantly lower than controls (AdSos = 1819.1 +/- 111.9 m s(-1); UBPS = 0.25 +/- 0.17), and BMD of hip (spine: 756.0 +/- 138.9 mg cm(-2); femoral neck: 583.6 +/- 61.8 mg cm(-2)) and vertebral (spine: 727.4 +/- 120.4 mg cm(-2); femoral neck: 592.8 +/- 82.1 mg cm(-2);) fracture was lower than controls (spine: 829.2 +/- 167.6 mg cm(-2); femoral neck 665.0 +/- 108.9 mg cm(-2)) (p < 0.001). Spine and femoral neck BMD significantly separated both types of fractures from controls, but AdSos and UBPS significantly separated only vertebral fractures. Femoral neck BMD has the best receiver operating characteristic (ROC) area for both hip (0.715, SE 0.043) and vertebral (0.693, SE 0.047) fractures, being significantly better (p < 0.01) than that of AdSos (0.503, SE 0.056) and UBPS (0.501, SE 0.057) for hip fractures, but not for vertebral fractures (AdSos 0.604, SE 0.050; UBPS 0.563, SE 0.048). In conclusion, QUS at hand phalanxes predicts vertebral fracture as effectively as BMD, but does not predict hip fracture.