We describe a single photon absorptiometric (SPA) technique, which enables differential estimation of the rates of loss from trabecular and cortical bone. Ten scans are obtained in the forearm: six in an area with about 7% trabecular bone and four scans in the adjacent distal area with a trabecular bone content of 25%. By comparing bone masses of these two sites in 19 postmenopausal and 53 premenopausal women, the postmenopausal trabecular bone loss was estimated to be approximately seven times greater than cortical loss within the first years of cessation of regular vaginal bleeding. On a group basis the bone loss at the distal forearm scan site (by SPA) corresponded closely to the spinal bone loss (by dual-photon absorptiometry). The reproducibility of the two scan sites in the forearm was 1-1.5% (CV%), which makes the method suitable for longitudinal studies. Corrections for variations in fatty tissue covering can be made without deterioration of the reproducibility. The high precision can only be achieved with a good calibration procedure; if calibration is not done the reproducibility error increases two- or threefold.