This study aims to analyze the classical mastery of mathematical reasoning abilities in the STAD learning model with the RME approach and the direct learning model. Another goal is to analyze the comparison of mathematical reasoning abilities with the STAD learning model and the RME approach with the direct learning model. This is a quantitative study with a quasi-experimental design in the form of a posttest-only control design. The population is class VIII students of SMPN 3 Mertoyudan. Using the cluster random sampling technique, two classes were obtained as research samples, with class VIII C consisting of 32 students as the experimental class and class VIII D consisting of 26 students as the control class. The data analysis technique used is the Z test and the independent sample T test. The test results with the Z test, in the experimental class and control class, is accepted, meaning that the two classes have completed classically. The test results with the T test in both classes, obtained then is rejected meaning that there is a difference in the mathematical reasoning abilities of the experimental class and control class students. After further analysis, the average value of the experimental class was while the average value of the control class was. Based on the results of the study, it can be concluded that: 1) the mathematical reasoning abilities of students who apply the STAD learning model with the RME approach achieve classical mastery; 2) the mathematical reasoning abilities of students who apply direct learning models achieve classical mastery; and 3) the mathematical reasoning abilities of students who apply the STAD learning model with the RME approach are better than students who apply the direct learning model.