Since bone mass has been shown to be an important determining factor of fractures in vitro, we undertook a study to evaluate whether bone mass measurements could separate postmenopausal women with vertebral compression fractures from women of a similar age without fractures. We also wanted to see if methods of measuring bone mass at the spine would be more sensitive or specific than methods that measured bone at the wrist or the entire skeleton. The techniques used were: total body calcium by neutron activation analysis (TBC), single photon absorptiometry (SPA), dual photon absorptiometry (DPA), and quantitative computed tomography (QCT). Normal women aged 20–85 were measured, but only those >50 yr were used in the analysis. Mean values for women with fractures were significantly lower than normals (p < .001): TBC 642 ± 103 g vs. 764 ± 114; SPA .658 ± .134 g/cm vs. .779 ± .142; DPA 3.75 ± .82 g/cm vs. 4.37 ± .86; QCT 59.0 ± 25.7 mg/cc vs. 92.6 ± 36.0. However, each of the methods showed considerable overlap between women with and without fractures. At 90% specificity the sensitivities of the tests were: TBC 34%; SPA 29%; DPA 33%; QCT 36%. When values were expressed as the % expected (based on age and height) then the sensitivities were: TBC 52%; SPA 36%; DPA 35%; QCT 44%. Using Bayes' theorem, we constructed curves showing the posttest probability of these tests at a prevalence of 20%. None of these bone mass measuring techniques showed complete separation between normal and osteoporotic women with fractures; about one-half of the women with fractures were below the normal range. The risk of having a fracture increases as bone mass declines, but our data suggest that bone mass is not the only factor leading to vertebral fractures in postmenopausal women.