In order to investigate a rhythmic movement ability of mentally retarded children, temporal accuracy of tapping task and stepping task were measured. On the tapping test, subjects were required to tap a key on the table exactly to sound signals being presented at a fixed time interval (900 msec). Being performed with their forefinger, this task was categorized as fine motor movement. In contrast, the stepping test was categorized as gross motor movement. Subjects were required to step on a given plate exactly to the sound signals. The tempo of the signals was the same as the one on the tapping test. The temporal accuracy of each test was assessed on the basis of timing error to the correct tempo from polygraphic records of 30 repetitive tappings or steppings. The mentally retarded subjects were 36 males (ages, 7-19; IQs, below 75). Their performances on both tests were compared with those of normal children (571 males, ages, 5-17). The metally retarded children at all age levels showed significantly greater error values than the comparative children. In order to compare the groups, the error values were transformed into T scores, the norm being constructed from the mean and SD of the nonretarded group of the same age and sex. On the tapping test, mean error score of retarded children was 60.9, being significantly greater than that of normal children (50.0). On the stepping test, the retarded children also showed a significantly greater mean error score (75.1) than the nonretarded (50.0). Consequently, the ratio of the tapping to the stepping test scores for the retarded group was 1.23. The results suggested that the retarded children lagged behind on the gross motor movement. Such tendency was observed in all age groups and was conspicuous in younger groups. The results were disscussed in terms of abilities of time perception, effector anticipation and effector efficiency.