Dual-energy X-ray absorptiometry (DXA) is now a commonly used method for the determination of bone mineral status and body composition in humans. The purposes of this study were to compare fat mass by in vitro neutron activation analysis (FMIVNA) with that by DXA (FMDXA) in an anthropometrically heterogeneous sample of healthy adult men (n = 33) and women (n = 36) (19 < or = BMI < or = 39), and to determine whether differences in fat mass estimates between the two methods (delta FM) were attributable to subject anthropometry as defined by several circumference (waist, iliac crest, thigh) and skinfold thickness (umbilical, suprailiac, abdominal) measurements. No significant differences between FMDXA and FMIVNA were observed in men (p = 0.46) or women (p = 0.09). The two methods were very highly correlated in both sexes (women r2 = 0.97, p < 0.001, men r2 = 0.91, p < 0.001), although the regression line for men was significantly different from the line of identity (p = 0.043). These results suggest modest trends toward underestimation of FMDXA in men when FMIVNA < 18 kg, and overestimation in men when FMIVNA > 18 kg. delta FM (IVNA-DXA) was not significantly related to any combination of skinfold thickness and circumferences in either gender. Age explained 27% of the variance in delta FM for the men (p = 0.008). Furthermore, delta FM was not significantly related to inter-method disparity in total-body bone mineral measurements in men or women (p < 0.05). The present study demonstrates strong correlation in fat measurements between IVNA and DXA in men and women ranging from normal to markedly obese. Correction for subject anthropometry does not significantly improve this relationship.