The low ability of social knowledge competence was caused by the less optimal use of instructional media during the learning process. It caused the students were difficult to understand the study. This study aimed to examine the effect of the Magic Box Media Assisted Rotating Trio Exchange (RTE) Learning Model Toward Social Science Competencies. This study was a quasi experiment study with non-equivalent control group design. The population of this study was 275 students. The sample was determined by cluster random sampling. The data were collected using test method with the usual multiple-choice instrument. As the result, it can get the data in the form of social knowledge competence value quantitatively that was analyzed using a t-test, namely polled variance. The result of data analysis obtained tcount=7,010 > ttable= 2,000 at the significant level of 5% (dks 31 +31-2-60) Therefore, HO was rejected and Ha was accepted so that there was a significantly different of the social knowledge competence between the group taught using the Rotating Trio Exchange assisted by Magic Box Media with the groups taught using conventional learning in fourth-grade of elementary school.