The main purpose of this research is to survey three different methods of variable practice in a skill learning. The research method was semi-experimental and the statistical population involves 100 subjects from Shahriyar primary schools of Karaj. They were randomly assigned to four 25 subject groups, 3 groups for variable practice programs and one as control. The 3 experimental groups were named A,B, and C group randomly. Diamond and weight of ball and distance from goal were manipulated in order to change practice parameters. Group A performed push pass kick from distances with ball size 4. Group B performed the kick from 15 distances with ball size 4, and group C performed the same as group B, but each of 20 kick with a different ball size, that is, 3, 4, and 5 sizes. Each subject had the opportunity of 60 trials per session, that is, 12 trials from 5 distances for group A. Group B had trials from 15 distances and kicked 4 kick from each distance. Group C performed 15 trials with one ball size, that is, 3, 4, and 4. The test was Mor-Christian Push Pass Test. Retention and transfer tests were given after 2 days. Significance was detected for the main effect of group, but for retention test only (a=0.95). No significant differences were found by the analysis of variance (ANOVA) for the main effect of group for transfer test (a=0.95).