This research is motivated by the learning process in Karangtengah elementary school teacher is still relatively conventional, especially ini fifth class is still low (48.4% of students have not reached KKM). The problem in this research are (1) the difference in student learning outcomes as mathematics students in learning using cooperative learning model type Two Stay Two Stray with students who are learning to use conventional learning; (2)?é?á the learning model cooperative type Two Stay Two Stray effectively used to improve learning outcomes in mathematics students? This type of research is quantitative. The population in this study were all fourth grade students of SD Negeri Karangtengah is 64 students by using saturated sampling . The data in this study using the test and non-test. This study design using True Experimental Design shaped Post Test-Only Control Design. Final data analysis using the average difference test and t test One Party Right. Based on the analysis of research data after getting treatment cooperative learning model type Two Stay Two Stray, an average of 81.355 experimental class learning outcomes and learning outcomes on average 57.848 control class. Results of test analysis the average difference in learning outcomes (experimental class and control class) coefficient t test of 2.667 with a significance level of 5% by interpolation obtained t table = 1.958 so that t > t table, then Ho is rejected and there are differences in learning outcomes among students using cooperative learning model type Two Stay Two Stray with conventional learning. From these results it can be concluded that cooperative learning model type Two Stay Two Stray effective and can improve the results of fifth grade students learn math SD N Karangtengah.