OBJECTIVE This study aimed to compare the accuracy of percent total body fat (%TBF) and percent segmental fat (%SF) measurements of new automatic single-frequency bioelectrical impedance analysis with 8 electrodes (SF-BIA8), referenced to the dual-energy X-ray absorptiometry (DXA), between different fatness level female groups. METHODS Subjects were 34 overweight females (OWG) and 24 normal females (NG). This study used the BIA analyzer of Tanita BC-118 which is a commercially available single frequency bioelectrical impedance analysis system with 8 tactile electrodes. RESULTS Test-retest reliability in each method was high (intra-class correlations > or =0.96). The SF-BIA8 tends to overestimate %TBF and %SF in the lower extremities and trunk compared with DXA, and consistency with DXA was especially low in %SFs in the lower extremities (left leg: r=0.475, SEE=4.673, right leg: r=0.494, SEE=4.772). Systematic errors regarding to the difference in fatness were found in %TBF and %SFs in lower extremities in OWG, and were found in %SF in upper extremities in NG. Furthermore, systematic errors regarding to the difference in body fat distribution were found in %SFs in lower extremities in OWG, and found in %SFs in upper extremities and trunk in NG. CONCLUSIONS SF-BIA can provide highly reproducible measurements of %TBF and %SF, but systematic errors based on the differences in fatness and body fat distribution were found. An improved prediction equation considering a variety in fat and body fat distribution will be needed in further investigations.