The purpose of the study was to cross-validate three different methods used in field studies for estimating body fat in a group of Danes (n = 139) aged 35-65 years with a wide range in body fat. On the first two-thirds of the participants (n = 93) multiple regressions equations were developed for body fat by means of height, weight, sex and age and either BMI, or by sum of four skinfolds or electrical impedance. A four-compartment-model based on measurements of both total body water (TBW) and potassium (TBK) was used as reference. On the remaining one-third of the subjects (n = 46) cross-validation of the developed equations was carried out. The multiple regression equation for impedance had higher R-square (R2 = 0.89) and lower residual error (SEE = 3.32 kg) than the multiple regression equations for skinfolds (R2 = 0.81, SEE = 3.91 kg) or body mass index (R2 = 0.85, SEE = 3.94 kg). The cross-validation study showed that all three simple methods give reliable average body fat estimates. The difference in estimates of body fat between the reference method and impedance, BMI and skinfolds was 0.76 kg, -0.64 kg and -0.21 kg respectively, none of these differences being different from zero. However, the variance of body fat estimated from BMI (4.84 kg) was higher than body fat estimated from impedance (4.36 kg) (P = 0.013). The multiple regression analysis further showed that measurements of skinfold do not seem to add significantly to the prediction of body fat if height and weight are included.(ABSTRACT TRUNCATED AT 250 WORDS)