Increasing body weight is associated both with higher bone mass and with lower rates of bone loss. Whether the effects of body weight are mediated by lean body mass (LBM) or fat body mass (FBM) is, however, uncertain because different studies have used different measures of bone mass and arrived at contradictory conclusions. The parameter actually measured is bone mineral content (BMC). Bone mineral density (BMD), bone mineral apparent density (BMAD), and the BMD/height attempt to "correct" BMC for differences in bone or body size, but these corrections may bias the analysis of the effects of body composition on the skeleton. To resolve this issue, we measured BMC at the total body, lumbar spine, proximal femur, and forearm using dual energy X-ray absorptiometry (DXA) in a population-based sample including 138 premenopausal women (age range 21-54 years, mean 35 years) and 213 postmenopausal women (age range 34-94 years, mean 68 years). BMD, BMAD, and BMD/ height were also calculated for each site. LBM and FBM were determined from the DXA whole body scan. In a multivariate analysis that included age and height, both LBM and FBM predicted total body BMC in pre- and postmenopausal women (p < 0.002 for LBM and FBM in both groups). LBM had a dominant effect on spine and forearm BMC in both groups (p < 0.004) and hip BMC in premenopausal women (p < 0.001), whereas both LBM and FBM predicted hip BMC in postmenopausal women (p < 0.001). However, as BMC was adjusted for bone or body size using BMD, BMAD, or BMD/height, FBM tended to become more important than LBM in the analysis. This was, in part, due to the fact that each of the correction factors in the BMD and BMAD calculations, as well as height, were highly correlated with LBM (r = 0.57 and 0.52 for height versus LBM in pre- and postmenopausal women, respectively [p < 0.001]), and weakly or not at all with FBM (r = 0.08 and 0.11, respectively). Therefore, dividing BMC by these correction factors tended to bias the analysis against potential effects of LBM on bone mass. Thus, the relationship between body composition and bone mass is critically dependent on which bone mass parameter is used in the analysis. Both LBM and FBM have important effects on bone mass, depending on the bone mass parameter used, the skeletal site measured, and menopausal status.