Eight untrained women (F), 13 untrained men (M), and 11 male bodybuilders (BB) did maximal elbow flexions on an isokinetic dynamometer at velocities of 30, 120, 180, 240, and 300 degrees/s, from which impact torque (IT), peak torque (PT), and work (W) were measured. Biceps and total flexor cross-sectional area (CSA) were measured by computerized tomographic scanning. Muscle fiber area, fiber composition, and collagen volume density were determined from single needle biopsies of biceps brachii. Biceps fiber number was estimated as the ratio of biceps CSA (corrected for connective tissue) to mean fiber area. PT and W decreased at higher velocities in M and BB but not in F; consequently, the correlation between CSA and PT and W was lower at 300 degrees/s (r = 0.58, 0.60) than 30 degrees/s (r = 0.80, 0.79). The ratio of PT to flexor CSA was similar in all groups at 30 degrees/s, whereas F had greater ratios than M and BB at the remaining velocities. F had greater W/CSA ratios than M and BB at all velocities. IT increased at higher velocities in all groups; the increase was greater in F and M than in BB. In contrast to PT and W, the correlation between IT and CSA was greater at 300 degrees/s (r = 0.67) than 30 degrees/s (r = 0.58), and there were no differences among groups in the IT/CSA ratios. Flexor CSA correlated negatively with the ratio of IT, PT, and W to CSA. Muscle fiber composition failed to correlate with any measure of strength. M and BB had greater biceps area, fiber number, and fiber area than F.(ABSTRACT TRUNCATED AT 250 WORDS)