Information is limited on the effect of exercise on bone density in young eumenorrheic athletes. We studied 12 Caucasian intercollegiate volleyball players (V), nine basketball players (B), ten swimmers (S), and 13 non-athletes (N) with bone density measurements by photon absorptiometry of their calcaneus and lumbar spine (L2-L4). The effect of athletic status on bone density was analyzed by multivariate analysis of covariance, with height and weight as covariates. The bone densities reported below are mean +/- SE, adjusted for the covariates; units = g.cm-2, P less than 0.005. The swimmers had a significantly lower mean density in the lumbar spine than all other groups; the non-athletes' mean density was also lower than that of volleyball players (V = 1.31 +/- 0.03, B = 1.26 +/- 0.04, N = 1.18 +/- 0.03, S = 1.05 +/- 0.03). The volleyball and basketball players' mean calcaneal densities were greater than those of the swimmers and non-athletes (V = 0.530 +/- 0.017, B = 0.564 +/- 0.023, N = 0.438 +/- 0.018, S = 0.375 +/- 0.019). The higher bone densities for athletes in vertical weight-bearing activities are consistent with some but not all published data. The swimmers' low bone density in the lumbar spine, less than published values for amenorrheic runners, was unexpected.