The objective was to determine the diagnostic sensitivity of spinal and femoral dual x-ray absorptiometry (DXA) and to study whether a combination of both sites may enhance discriminatory capability in regard to the presence of vertebral fractures. Spinal and femoral DXA were obtained in 324 postmenopausal women, of whom 90 had at least one vertebral fracture. Age-adjusted logistic regression analyses, ROC analyses, and sensitivity-specificity statistics were used to assess the discriminatory ability of spinal and femoral bone density (BMD) alone and in combination. The age-adjusted odds ratios per standard deviation decrease in BMD (OR) for spinal and femoral measurements were comparable (Ward's triangle: OR = 1.62; femoral neck: OR = 1.51; total hip: OR = 1.47; spine: OR = 1.34). Combining spinal and femoral bone density measurements did not improve diagnostic sensitivity of DXA considerably as compared to using BMD of a single site and adjusting the "fracture threshold." The conclusion drawn is that spinal and femoral BMD measurements using DXA have a comparable diagnostic sensitivity for vertebral fracture discrimination. Different individuals at risk for osteoporosis may be identified using both methods. The clinical usefulness of a combination of two bone density measurements needs further study in a prospective setting.