Résumé : Lumbar spine bone mineral density (BMD) was measured by dual-energy X-ray absorptiometry (DXA) (Hologic QDR 1000) and by 153Gd dual-photon absorptiometry (DPA) (Novo Lab 22a) in 120 postmenopausal women. Though a high correlation existed between the two techniques, the ratio between DXA and DPA values was not constant. Using DXA we observed a higher dependence of BMD on weight than in the DPA measurements. To investigate the different behaviour of DXA and DPA machines with weight, we analysed the effects of increasing thickness of soft tissue equivalents on the BMD of the Hologic spine phantom and on the BMD equivalent of an aluminium standard tube. Increasing tissue-equivalent thickness caused the phantom BMD measured by DPA to decrease significantly but had not effect on the DXA measurements. The different behaviour of DPA and DXA equipment with regard to the phantoms could account for the differences observed in the relations between BMD and weight in the patients. Using multiple regression we studied the influence of weight and body mass index on the relation between BMD measured by the two techniques. The introduction of either of these variables into the regression resulted in an improvement of the prediction of the DXA values from the DPA values. However, the residual standard error of the estimate was still higher than the combined precision errors of the two methods, so that no simple relation allows a conversion of BMDDPA into BMDDXA. Our results confirm that BMD is positively correlated with weight in postmenopausal women; the influence of weight on BMD is blunted when the Novo Lab 22a DPA machine is used for measuring bone mineral.