The objective of this study was to test whether differences between body fat percent (BF%) measured by densitometry and BF% predicted from body mass index (BMI) or bioimpedance (BIA) can be explained by differences in body build. Weight, height, sitting height (leg length), arm length, skeletal widths, BIA, bone mineral content from dual-energy X-ray absorptiometry, and BF% from densitometry were measured in 90 apparently healthy, young adult subjects (age range 18–31 years). The BMI was calculated and BF% predicted from BMI, age, and sex. BF% was also predicted from BIA. BF% measured by densitometry was 29.3 ± 5.0% for women and 14.6 ± 5.1% for men. BF% predicted from BIA (27.1 ± 6.7% in women, 19.0 ± 6.0% in men) was significantly different from measured values in both sexes. BF% predicted from BMI (25.9 ± 2.2% for women, 15.6 ± 2.4% for men) was only significantly different from BF% from densitometry in women, not in men. In both sexes skeletal widths, especially elbow width, were related to the prediction error of BF% from BMI, confirming the hypothesis that BF% predicted from BMI underestimates BF% in persons with a relatively slender body build or frame. The prediction error of BF% from BIA was found to be related to the length of the arms and legs, confirming the hypothesis that BIA overestimates BF% in persons with relatively long limbs.