Five computed tomography (CT) scans were taken at measured intervals of the legs and arms of young (n = 7) and elderly (n = 13) men. Cross-sectional areas (CSA) of the total limb, muscle plus bone and bone were measured in each scan, and skin plus subcutaneous tissue areas were calculated by subtraction. In addition, in the arm scans the CSA of the extensor and flexor compartments were measured, and in the leg the CSA of the plantar flexor compartment. A value for lean muscle within these compartments was calculated by excluding non-muscle tissue using density measurements based on Hounsfield units. Related volumes for the various components were also calculated using geometric formulae. The results showed that elderly limbs were of a similar overall size as the young, but elderly muscles were smaller (28-36%) with greater amounts of non-muscle tissue located within a muscle, particularly in the plantar flexors (81% more than in the young). Elderly arms had a greater amount of skin plus subcutaneous tissue than the young, but there was no difference in the legs. Muscle volumes were similar to in vitro results reported from cadaver studies and can be predicted from single mid-limb CT scans using regression equations. These results illustrate that, due to the substantially reduced amount of 'pure' muscle tissue in the elderly, comparisons of relative strength with other populations may be misleading unless appropriate measurements of muscle size are considered. Methods to estimate in vivo physiological CSA, which is considered the best means of normalizing strength, have been demonstrated in this study.