The purpose of this longitudinal study is to describe bone mass and body composition, and the annual changes in these measurements, among third grade students recruited from a suburban school district. Whole body bone mineral content (WBBMC), bone mineral density (WBBMD), fat, and lean mass were measured by dual-energy X-ray absorptiometry. Bone mass in the lumbar spine (LBMC) region of the whole body scan was also utilized. 773 students (38% white, 57% black, 5% other) had baseline visits; 561 had a second measurement a year later. At baseline, black children have significantly higher WBBMC, WBBMD, height, and lean mass than whites. Black males, but not black females, have a greater LBMC. There are no significant gender differences in body size, WBBMC, or WBBMD, although girls have a greater LBMC and fat mass, and boys have a higher lean mass. Most of these differences persist in visit 2. The annual change in bone and lean mass is greater in blacks. Stepwise linear regression analyses of bone mass on body size, gender, and ethnicity and their interactions indicate that log-transformed weight explains most of the variance in both WBBMC and WBBMD (multiple r2 = 0.90 and 0.64, respectively). There are significant black/white differences in intercepts and slopes. Other variables explain only another 1%-2% of the variance. The strongest Pearson correlations are between changes in bone mass and changes in lean mass and log-transformed weight (r ranging from 0.62 to 0.84, p = 0.0001). We conclude that there is a significant black/white, but not male/female difference in whole body bone mass and bone density before puberty. Ethnic and gender differences in bone and body composition suggest that the lean component may contribute to a greater peak bone mass in blacks vs. whites, and perhaps in males vs. females.