The loss of lean muscle mass is one of the hallmarks of protein-calorie malnutrition. Anthropometry is a standardized technique used to assess the response of muscle mass to nutrition therapy by quantifying the muscle and fat compartments. That technique does not accurately reflect actual limb composition, whereas computerized tomography does. Twenty lower extremities on randomly chosen men and women patients were evaluated by anthropometry and computerized tomography. Total area, muscle plus bone area, total volume, and muscle plus bone volume were correlated, using Heymsfield's equation and computerized tomography-generated areas. Anthropometrics overestimated total and muscle plus bone cross-sectional areas at almost every level. Anthropometry overestimated total area and total volume by 5% to 10% but overestimated muscle plus bone area and muscle plus bone volume by as much as 40%. Anthropometry, while easily performed and useful in large population groups for epidemiological studies, offers a poor assessment of lower extremity composition. On the other hand, computerized tomography is also easily performed and, while impractical for large population groups, does offer an accurate assessment of the lower extremity tissue compartments and is an instrument that might be used in research on lean muscle mass.