Background/purpose: The characteristics of each subject allow for adjustments to the learning model and the role that will be carried out by the teacher. One of them is Islamic education lessons with a diversity of student abilities and abstract and complex characteristics of Islamic education with its application. This research aiming to determine the improvement and difference in students' understanding of material due to the application of achievement grouping learning, and to examine teacher's role in the model. Materials/methods: This research is quantitative using a quasi-experiment approach. Grade VII students of Al-Irsyad Serang Junior High School in Banten became a sample. Through the purposive sampling technique, class A (experiment I) with the achievement grouping model, class B (control) with the conventional model, and C (experiment II) with the cooperative model. Data analysis was carried out using the Rasch model, general linear model test, independent sample T-test, Oneway Anova, and n-gain score. Results: The results of Anova test significantly showed that a difference in the comprehension value of each research class. The difference based on post hoc Anova test showed that the experimental class I was 18.74% superior to the control class, and 7.36% from the experimental class II. Conclusion: Furthermore, supported by the average n-gain score (0.5395), that the application of the achievement group learning model is able to increase students' comprehension scores with a moderate improvement category.