This research is motivated by the low ability of students' discussion due to learning that is still teacher-centered so that students only accept what is conveyed by the teacher which causes students to be less active in the learning process, one of which is in discussion. This is also caused by teachers not using learning strategies that attract students to be able to improve discussion skills. For this reason, an appropriate strategy is needed to be able to overcome these problems, by implementing the Teams Games Tournament (TGT) learning strategy. This research is a quasi-experimental research using quantitative data trends. The research design is a posttest-only-control design. The population in this study amounted to 123 students consisting of five classes. The sample for this study used 2 classes, namely class VII.5 as the experimental class and class VII.4 as the control class. The results showed that the average discussion ability of the experimental class was 73.73 and the average discussion ability of the control class was 49.48, so it can be seen that the average experimental class was greater than the control class with a difference of 24.25. Based on the sig (2-tailed) data from the data obtained, namely 0.000, it can be concluded that the value of 0.000 <0.05, it can be concluded that Ho is rejected and Hi is accepted or there is a significant influence on students' discussion abilities in the experimental class and control class.