Associations of body weight and body composition with bone mineral density (BMD) were examined in 261 postmenopausal women. BMD, body fat, and body nonfat soft tissue (NFST) were measured by dual-energy X-ray absorptiometry (DXA). A height-independent BMD variable (HIBMD) was calculated to correct for differences among individuals in bone thickness, a dimension that is ignored by DXA scanners. HIBMD was calculated as BMD divided by height at the spine and femoral neck, and BMD divided by the square root of height at the total body. Weight, fat, and nonfat soft tissue were all positively correlated with both BMD and HIBMD, but the magnitudes of regression and correlation coefficients were lower when HIBMD was the dependent variable. The weight-independent associations of body composition with HIBMD were examined by including weight and % NFST together in linear models. In these analyses, weight was positively associated with HIBMD at all three skeletal sites (r = 0.22-0.26, P < 0.05), % NFST was not associated with HIBMD at the spine or femoral neck (r = 0.01-0.02), and there was only a weak inverse correlation of % NFST with total body BMD (r = -0.12, P < 0.05). These findings are consistent with those of previous studies demonstrating positive associations between body weight and BMD. In addition, they demonstrate that once bone thickness and body weight are taken into account, body composition appears to have little if any independent effect on bone density at the skeletal sites measured. This finding is consistent with the hypothesis that the protective effect of body weight is brought about predominantly through its mechanical force on the skeleton.