We compared different methods of bone densitometry in women with spinal osteoporosis and normal subjects to assess their discriminatory capability. The methods used included: quantitative computed tomography of the spine (QCT) specified as to trabecular (QCTtrab) and cortical bone (QCTcort), dual-photon absorptiometry of the spine (DPAspine), single-photon absorptiometry of the distal and proximal forearm (SPAdist and SPAprox), and quantitative roentgen microdensitometry of the phalanx (QMD). A total of 25 postmenopausal osteoporotic women and 24 healthy comparison subjects matched for age and years since menopause were studied. In the osteoporotic group an average decrement of the axial bone mineral density of -50% (p less than 0.001) and -20% (p less than 0.001) were observed for QCTtrab and QCTcort, respectively. For DPAspine, SPAdist, SPAprox, and QMD the difference between normal and osteoporotic subjects was -20% (p less than 0.001), -12% (p less than 0.05), -7% (NS), and -6% (NS), respectively. With the peripheral measurements (SPA and QMD), alone or in combination, no adequate discrimination between women with or without vertebral compression fractures could be obtained. Although QCTtrab showed the highest diagnostic sensitivity (81%), it appears not to be superior to DPAspine. Combinations of the various axial and peripheral measurements did not result in an essentially better sensitivity. In normal women as well as in osteoporotic individuals the trabecular and cortical QCT measurements showed two opposite trends, suggesting an increase in cortical and a decrease in trabecular density from L1 to L3.