Between May 1990 and February 1991 we enrolled 212 patients of 410 referred to our clinic for bone mass evaluation according to different diagnosis or therapeutic procedures. These patients were divided into 7 groups. The question had to be answered, which method of measurement and which skeletal site would give better information about the influence of the diseases of therapeutic procedures. We measured the areal mineral density of the lumbar spine with dual x-ray absorptiometry (DEXA) and the mineral content of the total and trabecular bone mass of a cross-sectional slice at the distal radius with peripheral quantitative computed tomography (pQCT). The results were statistically analysed. We calculated the percentage deviation from young normals. Trabecular mineral content showed the greatest difference (18% to 44%), compared to the healthy reference group, corresponding to the known higher turnover of this bone compartment. In contrast, the difference of the areal density of the lumbar spine was found to be lowest (4% to 27%). In order to compare the methods objectively, we additionally calculated a ROC-analysis of two postmenopausal groups (51 healthy, 46 osteoporotic) to demonstrate the diagnostic validity. The validity proved to be substantially better when purely trabecular bone at the distal radius was measured. The ROC-curves showed a sensitivity of 84% for radial trabecular bone on a specificity level of 90%, whereas the lumbar spine values presented with a sensitivity of only 68%.