The purpose of this study was to compare the estimates of lean body mass (LBM) and percent body fat (%BF), as predicted by bioelectrical impedance (BIA) and sum of skinfolds (SF), with those derived by hydrostatic weighing (HW) obtained before and after a 10-wk diet and exercise regimen. The experimental (E) group consisted of 17 healthy male subjects; 20 healthy males served as the control (C) group. Post hoc Scheffe contrasts computed on E group data indicated that, for both LBM and %BF, the Lukaski and Segal BIA equations, as well as the Durnin SF equation, derived mean values that were not significantly different (0.05 significance level) from HW in both pre- and postregimen conditions. For LBM, the same equations derived the following significant (P less than 0.01) correlation coefficients for both pre- and postregimen data: Lukaski, 0.87 and 0.85; Segal, 0.89 and 0.87; and Durnin, 0.90 and 0.88. For %BF, the correlation coefficients were slightly lower but remained statistically significant (P less than 0.01). The findings of this study suggest that the BIA method, by use of either the Lukaski or Segal prediction equations, is a valid means of predicting changes in human body composition as measured by the Siri transformation of body density.