In order to assess precision of peripheral quantitative computed tomography (pQCT), duplicate bone mineral density (BMD) measurements at the radius were performed in 20 healthy premenopausal, 20 healthy postmenopausal, and 20 osteoporotic postmenopausal women using a Stratec XCT-960 system. The short-term reproductibility in vivo for the total, trabecular, and cortical regions of interest (ROI) was expressed as the absolute precision error (standard deviation, SD) and as the relative precision error (SD/mean x 100, or coefficient of variation, CV, in %). Reproducibility in vivo was good in all volunteers but was influenced by the study group and the ROI. The precision error for trabecular BMD was 3 mg/cm3, or about 1.6%. This is large relative to the aging decrease of 0.22%/year, or to the difference (12 mg/cm3, or 7%) between osteoporotic women and postmenopausal controls. In order to compare pQCT to dual X-ray absorptiometry (DXA) at the forearm and at the lumbar spine (L1-L4), 40 premenopausal healthy controls, 40 postmenopausal healthy controls, and 35 postmenopausal osteoporotic women were assessed. DXA measurements performed at the ultradistal, middistal, 1/3, and total ROI of the radius showed only moderate correlations between r = 0.38--0.75, r = 0.27--0.64, and r = 0.38--0.53 for the comparison versus pQCT total BMD, versus pQCT trabecular BMD, and versus pQCT cortical BMD, respectively. Correlations of DXA at the lumbar spine and pQCT were between r = 0.18 and 0.44. DXA at radius and spine was able to discriminate between post- menopausal controls and osteoporotic women (p = 0.001--.004),but BMD measurements by pQCT did not show this ability (p = 0.15--0.52). However, two nonstandard pQCT parameters, namely the surface area of the cortical bone and the cortical BMC were factors that discriminated well between these two groups (p = 0.002, p = 0.005, respectively). These pQCT parameters also yielded the highest relative annual changes in pre- and post-menopausal control subjects. The measurement of cortical bone in the distal radius proved to be a good predictor of vertebral fracture status and was a good indicator of age-related skeletal change. Our data emphasize the importance of cortical measurements when using pQCT of the radius to assess osteoporosis.