Previous radiographic methods for the in vivo quantitation of bone mineral have generally expressed the radiographic density of a scan path across bone in terms of the thickness of reference material which gives the same attenuation of the same heterogeneous x-ray beam (4). The attenuation of the x-ray beam under such conditions is related to the amount of soft tissue over the bone and to the amount of cortical and trabecular bone present. Thus, such methods virtually ignore the relatively minor contribution of the trabecular pattern per se to the overall attenuation of the x-ray beam. In addition, x-ray fields sufficiently large to include the reference wedge add large amounts of x-ray scatter as an additional variable. Known data demonstrate that trabecular bone is more metabolically active than is cortical bone, in terms of mineral turnover rate (1, 3). Since vertebral trabecular bone has been shown to be the most informative to study, with regard to bone mineral metabolism (7), a method has been developed which is designed to permit quantitation of the changes in trabecular architecture which occur in the vertebrae of intact patients. This method, because it does not need a reference wedge, permits the utilization of films made with field sizes no larger than a vertebral body, thus eliminating most x-ray scatter. The method is based upon the hypothesis that vertical orientation of trabeculae is uniform when a vertebral body is viewed from the front or side, but the orientation of the trabeculae is nonuniform when the vertebral body is viewed from the top. This hypothesis appears reasonable, on the basis of the roentgenographic appearance of the trabeculae of excised vertebrae made from the side and top and also on the basis of the pathologic aspect of vertebral-body horizontal and vertical sections (6). The method is also based upon the hypothesis that it is possible to superimpose on a raw densitometric scan a continuously computed, weighted mean line along which the “width” of trabeculae can be determined and which, because it follows the regional density changes, is relatively independent of the regional alterations in optical density in any given roentgenogram. Methods A scanning microdensitometer,3 built for the specific research requirements of this laboratory and de-scribed previously (5), is being utilized for analysis of the roentgenograms. The output signals from the densitometer are fed simultaneously into an analog-to-digital converter,4 into a single channel strip-chart recorder,5 and into two TR-20 analog computers.6 The single channel strip-chart recorder displays the raw densitometric scan information. The analog computers have been programed so that the raw densitometric scan is smoothed, removing the high frequency fluctuations.