A variety of techniques for assessing human body composition have been developed. However, the validity of these methods has not been compared in the same subjects. The purposes of this study were (1) to evaluate the validity of %body fat using 5 kinds of indirect method-4 brands of bioelectrical impedance (BI) analyzers and skinfold thickness-against body fat data derived by underwater weighting, and (2) to discuss the accuracy of each method. The subjects were 40 men and 61 women, aged 20 to 67 years (men : 31.0±11.0yr, women : 36.0±13.0yr). The BI analyzers used for this study were the Sekisui SS-103, Omron HBF-300, Ueda TYPE-PS, and Tanita TBF-534, and an Eiken caliper was used for skinfold thickness detrmination. Percentage of body fat etimated by each method was significantly correlated (r=0.55-0.77) with the data from underwater weighing in both sexes, and the standard error of estimate (SEE) for %body fat was about 3% for men. However, %fat estimated by the HBF-300 and skinfold thickness for women were significantly different from the results of underwater weighing, giving a SEE of 4% or over for all methods. The accuracy of %fat estimated by the TBF-534 was slightly worse than that for the other BI analyzers in both sexes. In women, the accuracy of %fat estimated by skinfold thickness was slightly worse than that by the BI methods. From these results, it is concluded that %fat can be estimated by each method without a large discrepancy. However, the validity of each method for estimating %fat may not be as high as other reference methods. Further research is necessary for improvement of prediction accuracy.