The research aimed to investigate effect of Kabaddi local game reducing the percentage and severity of stuttering among girls 7 to 11 years old in Saravan city. The population consisted of 23 students from Saravan city primary schools in the 2016-2015 school years, of which, 20 as sample size using non-random sampling method selected purposefully. The method used in this study is quasi-experimental and pre-test - post-test control group so that effect 8 weeks Kabaddi game in both group intensity and frequency of symptoms of stutter were measured. To measure the frequency and severity of symptoms of childhood stuttering, stuttering tests and tests to measure severity of stuttering were used. First test Kolmogorov - Smirnov test to determine the normality of data distribution and equal variances was used to examine Levine. Then, to test hypotheses, parametric t-paired test to determine intergroup differences between pre-test and posttest groups and parametric t-independent test to determine differences between groups in pre-test and post-test were used. After running 8 weeks in three sessions per week Kabaddi game for half an hour each time frequency and severity of symptoms stutter were measured again. The results showed that children with stuttering in two training groups and controlling for age, severity and the percentage of stuttered were homogeneous and uniform because the comparison mentioned in the pre-test variables were not significant. Difference between mean changes severity and the percentage of children stutter that exercise in control groups was not statistically significant.