Both fetal and neonatal nutrition may influence the body fat content of infants. Epidemiologic field studies would be facilitated by a simple method of measuring total body fat (TBF). The accuracy of a method, based on skinfold measurements, of estimating TBF was evaluated in 22 infants by comparing the results with those obtained by the body water dilution technique. Because the accuracy was poor, a modification was evaluated in 14 of the infants. The measurements were used to calculate the amounts of subcutaneous and nonsubcutaneous body fat. Estimates of the thickness of subcutaneous adipose tissue made with calipers and ultrasonography were compared with each other and with those obtained with Futrex 5000, a device based on the principle of near-infrared interactance. The composition of the adipose tissue in biopsy specimen from 38 infants was analyzed and contained a mean of 0.66 g of fat/cm3. Results obtained by Futrex 5000 correlated with subcutaneous adipose tissue thickness only when the latter was thin. Estimates of adipose tissue thickness by calipers were significantly higher than estimates by ultrasonography at the forearm, thigh, and calf, whereas the opposite was found at the triceps, biceps, umbilicus, and nipple. The amount of nonsubcutaneous fat, in relation to body weight minus subcutaneous fat, increased with age and was higher in girls than in boys. Skinfolds were poor predictors of TBF. However, it may be possible to predict TBF from anthropometric measurements if appropriate knowledge about the growth and development of adipose tissue in infancy are available.