Abstract The maximal speeds of discrete arm and leg movements similar to the actions of throwing a baseball or kicking a football were tested in 80 college men. Repetitive (cyclic) movements involving both arms or both legs were also measured. Individual differences were highly reliable, ranging from r = .92 to .99. Intercorrelations between discrete and repetitive movements were very low, ranging from r = .03 to .16. While the correlations were somewhat higher between arm speed and leg speed, namely .39, the amount of task specificity under the most favorable circumstances was about 85 percent, compared with 15 percent common variance that could be ascribed to a general “speed of limb movement” factor. The results suggest the need to revise current concepts of the nature of motor abilities, since individual differences in making a fast movement are evidently quite specific to the particular motor task, rather than existing as a speed component that could be measured in a motor ability test.