Based on the result of interviews and observations, I found the problem that became the reason for this research. One of that is the learning model commonly used in Muhammadiyah Junior High School 1 Yogyakarta (SMP Muhammadiyah 1 Yogyakarta) has not been optimal to create mathematics learning interaction of students. This research aims to determine (1) the presence or absence of difference in mathematics learning interaction between a student is taught using cooperative learning model Think Pair Share (TPS) type with students taught using direct learning model and (2) which learning model is more effective between cooperative learning model Think Pair Share (TPS) type with direct learning model toward mathematics learning interaction of students. The type of this research is quantitative research, using experimental design. The population in this research was the students of regular class VII in SMP Muhammadiyah 1 Yogyakarta, even semester academic year 2016/2017. Research sampling done by random sampling toward class, selected VII F class as the experimental class, and VII G class as the control class. Technique data collection using the observation method. The prerequisite analysis test used a normality test using the Chi-square formula and homogeneity test using the Bartlett test. Data analysis used a hypothesis test using a t-test with a significant 5% level and degrees of freedom 61. Based on the results of the hypothesis, I test obtained  and . Because , so  is accepted. So, concluded that there is no difference in mathematics learning interaction between students taught using cooperative learning model Think Pair Share (TPS) type with students taught using a direct learning model. Because of the hypothesis, I obtained no difference, and then the hypothesis II test is not needed to continue. So, concluded that the cooperative learning model Think Pair Share (TPS) type is not more effective than the direct learning model.