The purpose of this study was to re-examine the effects of the practice variability in motor learning. The practice variability effects were analysed also with reference to two viewpoints: one is whether test target is contained in practice or not, the other id the test-task difficulty. Subjects were 120 ten years old elementary school children. The task administered was to hit the tennis ball rolled from about three meter away with a field hockey stick aiming at the targets which were placed five meter away from hitting point. The targets were set at the angles of ten to ninety degrees from the ball rolling direction. Numbers of learning trials were eighty, and two test blocks consisted of twenty trials each. Subjects were devided into four groups according to learning condition: 1) one-target (test target) group 2) one-target (non-test target) group 3) four-target (test target contained) group 4) four-target (non-test target) group The results were as follows: 1. Four variability groups were superior to one target groups. Especially, in the case of easier task, this tendency was more evident. 2. Test target practice groups seemed to get higher score than non-experience groups . But there were no significant differences between them. 3. When two variability groups were compared, there was no difference between test target contained group and test target uncontained group. These results support the schema theory. But, about the formation of motor response schema, this study could not make clear the role of test target practice.