This study estimated total body water (TBW) in four groups (twelve per group; sedentary and highly trained men and women) at the time of 2H dosing (T0) and after a 3·5 h equilibration period (Teq). Standard TBW calculations were employed at T0 (no correction for disproportionate urinary tracer loss) and Teq (correction for urinary tracer loss only), plus those calculations that corrected for a disproportionate urinary tracer loss and insensible tracer loss respectively. The measurement of body density enabled the four TBW estimates to be compared for the determination of three-compartment % body fat (BF). The very small difference between the standard and corrected T0 TBW data was not significant (P=0·914) and no Group×TBW interaction was identified (P=0·125). These results reflect the closeness of the 2H concentration in the urine produced during the equilibration period and the Teq saliva samples. The associated mean % BF values were essentially identical. Although correcting for insensible 2H losses in addition to urinary losses at Teq produced a statistically significant (P<0·001) lower mean TBW (about 200 g) than the standard calculation, this translated to a small difference in % BF (0·3). The larger difference (about 500 g, P<0·001) between the two (T0, Teq) corrected TBW calculations was also associated with a small body composition difference (0·1 % BF), which was less than the propagated error (0·3 % BF) for the three-compartment body composition model. Corrections to the standard calculations of TBW at T0 and Teq for a protocol employing a brief equilibration period (3·5 h) were therefore of marginal use for improving the accuracy of % BF estimates. The TBW difference over time (T0v. Teq) also had little impact on % BF values.