Total body mass is a major determinant of bone mass, but studies of the relative contributions of lean mass (LM) and fat mass (FM) to bone mass have yielded conflicting results. This is likely because of the use of bone measures that are not adequately adjusted for body size and, therefore, not appropriate for analyses related to body composition, which is also correlated with body size. We examined the relationship between body composition and peak bone mass in premenopausal women aged 18-30 yr using both size-dependent and size-adjusted measures of bone density and body composition, as well as statistical models adjusted for size-related factors. We measured total bone mass and areal bone density using dual-energy X-ray absorptiometry, and used established formulas to calculate estimates of volumetric (size-adjusted) bone density. LM tended to be positively associated with bone both before and after adjustment for size-related factors. FM and body fat percentage, however, were positively associated with size-dependent bone measures, but adjusting for size removed or reversed this association. These findings suggest that the association between bone mass and body composition, especially FM, is dependent on the bone measures analyzed, and that determining the most appropriate size-adjustment techniques is critical for understanding this relationship.