One hundred and four female and male athletes underwent determinations of hydrodensitometry (HD) and tetrapolar bioelectrical impedance (TBI) under uncontrolled (measurements made without regard to preceding exercise, level of hydration, or eating) and controlled conditions (measurements made two hours after consuming a light meal and no preceding exercise). Fat-free mass (FFM) was estimated from TBI measurements using a previously determined multiple regression equation. No difference (p greater than 0.05) was found in mean FFM values by method or experimental condition. In the uncontrolled state, fat mass and percent body fat (%BF) by impedance were lower (p less than 0.05) than that observed by HD or by TBI under controlled conditions. Linear regression analyses between FFM determined by HD and predicted by TBI under controlled and uncontrolled conditions yielded regression coefficients similar to the line of identify. The relationship between %BF by HD and calculated from impedance-predicted FFM and body mass under uncontrolled conditions was different (p less than 0.05) from the line of identify. In contrast, the regression of densitometric against impedance-predicted %BF under controlled conditions was similar to the line of identify with an error of 2.8%. These findings indicate the need for controlled measurement conditions to obtain valid body composition estimates using the TBI method in healthy people.