Reliability of B-mode ultrasound for the measurement of body fat and muscle thickness.

B-mode ultrasound was used to measure fat and muscle thicknesses on 30 subjects (17 men, 13 women, age = 20-37 yr) at 14 sites (triceps, biceps, forearm, chest [males only], subscapular, axilla, abdomen, suprailium, lumbar, quadriceps, suprapatellar, hamstrings, medial calf, and posterior calf) on two different days. Quadruplicate photographic images (trials) were printed from a single measurement at each site on each day. Two investigators each measured two of the images from each site. Each thickness was measured to the nearest 0.05 mm with a vernier caliper. Generalizability theory was used to determine the relative contribution of subjects, investigators, days, and trials to the total measurement variability. Subjects accounted for 84-96% of the variance in the muscle measurements and for 79-97% of the variance in the fat measurements. A subjects-by-day interaction accounted for 2-13% of the variance in muscle measurements and 2-12% of the variance in fat measurements. The contribution by investigators and trials to the variance was less than 1%. Generalizability coefficients (G) exceeding 0.92 were obtained for all sites for muscle measurements, while G for fat measurements exceeded 0.90 for all but the axilla site (G = 0.88). These results indicate that B-mode ultrasound is a highly reliable method for the measurement of both fat and muscle thicknesses in young males and females. © 1992 Wiley-Liss, Inc.