Abstract: This research aims at finding out: (1) which one results better mathematics learning achievement among type SNH through RME approach, type TSTS through RME approach, or direct instruction; (2) which one has better mathematics learning achievement among students with high, average, or low AQ; (3) at each of types (SNH through RME approach, TSTS through RME approach, and direct instruction), which group of studentsÂ has better mathematics learning achievement among groups with high, average, or low AQ; (4) at each of AQ categories (high, average, and low), which one results better mathematics learning achievement among type SNH through RME approach, type TSTS through RME approach, or direct instruction.This research employed quasi-experimental design taking all of the seventh grade students of State Junior High Schools in Surakarta as the population. The sampling technique used was stratified cluster random sampling. There were 285 students selected as the sample of this research. The techniques of collecting data were documentation which was used to know the studentsâ mid-test score in the even semester as the prior knowledge score, test which was used to know the studentsâ mathematics learning achievement, and questionnaire which was used to know the studentsâ AQ category. The hypothesis test used 3x3 two-way analysis of variance with unbalanced cells in the level of significance 0.05. The conclusions of the research were as follows. (1) Cooperative learning type SNH through RME approach results better mathematics learning achievement than type TSTS through RME approach does and cooperative learning type TSTS through RME approach results better mathematics learning achievement than direct instruction does; (2) Students with high AQ have better mathematics learning achievement than those with average and low AQ, and students with average AQ have better mathematics learning achievement than those with low AQ; (3) In the use of cooperative learning model type SNH through RME approach, students with high AQ have better mathematics learning achievement than those with low AQ, students with high AQ have relatively the same mathematics learning achievement as those with average AQ, and students with average AQ have relatively the same mathematics learning achievement as those with low AQ. In the use of cooperative leaning model type TSTS through RME approach, students with high, average, and low AQ have the same mathematics learning achievement; (4) For studentsÂ with high AQ, the use of learning model type SNH through RME approach results better mathematics learning achievement than that of type TSTS through RME approach and direct instruction, and the use of learning model type TSTS through RME approach results the same mathematics learning achievement as that of direct instruction. For students with average and low AQ, the use of learning model type SNH through RME approach results the same mathematics learning achievement as that of type TSTS through RME approach and direct instruction, and the use of learning model type TSTS through RME approach results the same mathematics learning achievement as that of direct instruction.Key words: Cooperative Learning Model, Structured Numbered Heads, Two Stay Two Stray, Realistic Mathematics Education Approach, Adversity Quotient