This study aims to find out if the results of students' learning through the implementation of a group investigation learning model through mathematical domino media is better than the implementation of a conventional learning model on the rank number and root form materials. The study was a quasi-experiment with the design of a randomized control group pretest-posttest. The population of this research is the entire class IX SMP Negeri 1 Lhokseumawe consisting of 9 classes. The research sample was determined by a randomized technique that made 2 classes namely experiment class (IXA) and control class (IXB). The results of the student posttest data test show that the data is distributed normally and homogeneously, so it can be analyzed with one-side t-test statistics at a significant [alpha]= 0.05. Based on the results of data processing against the posttest students obtained sig scores. (2-tailed) posttest data is 0.001 which means less than [alpha]= 0.05, this means H[subscript 0] is rejected, so it can be concluded that the learning results of students taught by the group investigation learning model through domino mathematics media are better than conventional learning models on the rank number and root form materials in grade IX of SMP Negeri 1 Lhokseumawe.