This study investigated the validity of Fitts' index of difficulty as a measure of task difficulty perceived by the performer. Eight subjects performed self-terminated horizontal elbow-extension movements toward targets of three indices of difficulty in two task conditions. In the target-size condition, triceps and biceps activities during acceleration and deceleration respectively decreased with increasing index of difficulty, resulting in a lower peak velocity and longer movement time. In the movement amplitude condition, however, triceps activity after movement onset and biceps activity during deceleration increased with increasing index of difficulty, resulting in a higher peak velocity. The movement time also increased as a function of the index of difficulty given increase in the distance traveled by the limb. These results suggest that Fitts' index of difficulty may not reflect the difficulty of task performance because its effect on the pattern of motor output is dependent upon the task variable manipulated in measuring the index of difficulty.