In the present research, the peers of children with autism at primary school level and in an inclusive environment were taught using direct modeling and video modeling education processes, and it was observed whether or not they could effectively and efficiently teach how to play games to their friends with autism. This study used adapted alternating treatments design from single subject designs. The research included 21 students from the first and second grades of primary education, 18 of whom participated as peer tutors with normal development, and 3 of whom participated as peer tutees with autism. The dependent variable of the research was the game learning skills determined through interviews with teachers for each sample. The independent variables were the implementation of direct modeling education and video modeling education by peers. The effectuality results of the study indicated that the participants played these games at 83-100% accuracy level and generalized it to different environments. In addition, the participants were observed to have the rules of games at 83-100% level one and three weeks after the completion of the implementation. Comparing the efficiency of the two education implementations, no significant difference was observed between direct modeling and video modeling education implementations in terms of the efficiency variable. The social validity findings of the research indicated the positive expressions of peer tutors regarding the study.