<p><em>This research is aimed to know</em><em>: (1) Whether the increasing of mathematical learning result of student through problem-based learning model is higher than the improvement of mathematical learning result of students who do not get problem based learning model (2) How the interaction between learning and early math  ability of student mathematical learning result. This study was a quasi-experimental study. The population of this study is the entire class of five semester mathematics education study which amounted to approximately 184 people. Randomly, two classes were selected from six classes. The experimental class is treated with problem-based learning model and control class is not given problem-based learning treatment. The instrument used is a description test. The data analysis is done by analysis of two-way variance (ANAVA). The results showed (1) improvement of student learning outcomes through problem-based learning model is higher than the increase of learning outcomes in students who are not treated problem-based learning model (control class). (2) There is no Interaction between learning and early math ability of student mathematical learning result.</em></p>