This study investigated the effect of the realistic mathematic education (RME) learning model, compared with the conventional model on mathematics learning outcomes in social arithmetic material in class VII SMP Al-Furqon Pejagan Jambesari Bondowoso. It uses a quasi-experimental design in the form of a nonequivalent pre-test and post-test. The population in this study were all seventh-grade students of SMP Al-Furqon Pejagan Jambesari Bondowoso, represented by two samples; class VII A as the experimental class and class VII B as the control class. The paired sample t-test resulted in a sig value in pair 1 of 0.000 < 0.05 and pair 2 of 0.289 > 0.05. It implies that the realistic mathematics education (RME) learning model affects the improvement of the student learning outcomes (pair 1), while the conventional learning model does not affect the student learning outcomes (pair 2).