The purpose of this study was to obtain empirical evidence about the use of cooperative models of Team Games Tournament to increase the ability of students on solving problems with the summation material fractions. To achieve these objectives, the research carried out in the form of an experiment by comparing the problem solving ability of students to the material sum of fractions through the use cooperative model of TGT and students who received conventional learning. The design is a pretest-posttest control group design. The sampling technique used is purposive sampling technique. The instrument used is to use tests that pretest and posttest. The data were analyzed quantitatively for the results of the pretest, posttest, and normalized gain value. Based on data analysis in this study we concluded that there are differences in problem solving ability of students to the material sum of fractions through the use of cooperative models of Team Games Tournament with students who studied with conventional models, and improved problem solving abilities of students in the material that follows the fractional summation cooperative learning of TGT better than students who take the conventional learning model. Therefore, the ability of solving problems of students at grade material fractions summation cooperative modeled of TGT has increased quite good.