Body composition is an important indicator of nutritional status. The most commonly used indirect method for estimating body fat is based on measurements of subcutaneous fat tissue. It has been suggested that ultrasonic measurements may be more precise than those of the caliper and therefore may yield more accurate measures of subcutaneous fat tissue. This study was designed to correlate ultrasonic and caliper measurements of subcutaneous fat with body density determined by hydrostatic weighing. Subcutaneous fat thickness was measured at seven body sites (triceps, biceps, subscapula, waist, suprailiac, thigh, and calf) with a Lange skinfold caliper and an ADR ultrasonic scanner, equipped with a display-screen, 7MHz transducer, and electronic calipers. Regression equations to predict body density, and hence body fat, were derived for each technique using a minimal number of body sites. The sample consisted of 124 white men, aged 18 to 30 yr. Mean body density determined by hydrostatic weighing was 1.07 g/ml (SD +/- 0.01) and mean body fat was 12.7% (SD +/- 5.8). Both ultrasonic and caliper measurements of waist, thigh, and triceps had the highest correlation with body density. Regression equations using these three sites in all possible two-site combinations were derived for each technique. The predictions of body density from these equations did not differ significantly. These results suggest that in free-living, nonobese, white men, body fat can be estimated with nearly the same degree of accuracy using either the caliper or ultrasonic technique.