The variability in measuring total body water (TBW) using deuterium oxide (2H2O) dilution has not been extensively tested. The purpose of this study was to determine the variability of weekly measurements of TBW utilizing the 2H2O dilution technique in 10 males and to determine the amount of biological versus technical variability using a generalizability ANOVA (GENOVA). Ten male subjects aged 21-32 yr were tested on four separate occasions for body weight (BW), underwater weight (UwWt), and TBW. BW remained quite stable across the four sessions with a 0.7% coefficient of variation (CV). Both UwWt and TBW had a CV of 4% across the four sessions. No differences were found between TBW estimates from plasma, saliva, or urine. GENOVA analysis determined that 25-50% of the total variability in measuring TBW was attributed to the sample analysis/separation procedures. Four percent of the TBW in these subjects was 1.8 l. This technique would therefore not be expected to detect changes less than 0.9 l (50% of 1.8 l). The estimates of %FAT from TBW were lower but not significantly different from UwWt. Blood, urine, and saliva were able to produce similarly repeatable measures of TBW.