To identify the factors associated with greater bone density among athletic individuals, we recruited three distinct groups of young male subjects. Twenty were nationally ranked water polo players, 19 were engaged in weight-training programs, and 20 subjects comprised a nonexercising comparison group. All participants had measurements of spinal trabecular and integral bone density by quantitative computed tomography as well as a determination of hip bone density by dual photon absorptiometry. A series of potential predictor variables included maximal O2 uptake, back strength, leg strength, total kilocalories expended per day, body mass index, paraspinous muscle cross-sectional area, percent body fat, daily calcium intake, and age. We found no significant differences for any of the bone density measures between the two groups of athletic subjects, whereas bone density was generally significantly lower among the nonexercisers compared with either exercise group. Correlation analysis found only weak and somewhat inconsistent relationships when each of the subgroups was examined separately; however, when all subjects were assessed collectively, many more correlations reached significance. Paraspinous muscular area was found to be most robust in this regard, being significantly correlated with all three bone density measures (r = 0.33-0.55). By using step-wise regression analysis in each subgroup, we observed a consistent significant contribution (R2 = 0.18-0.44) of paraspinous muscle area to the variability in bone density at the spine and the hip. When the data of all three subgroups were pooled, regression analysis reconfirmed the importance of the muscle parameter (R2 = 0.06-0.27) to bone density variation, but more importantly it showed that differentiation based on exercise status was most significant (R2 = 0.18-0.22).