This research motivited by the low of students' mathematical problem solving abilities. The one of alternative learning thatis applied to improve the mathematical problem solving ability of students by using Open-ended approach. The purpose ofthis research was to find out whether there were differences in the improvement of mathematical problem solving abilitiesbetween students using Open-ended approach and students used the Problem-Based Learning model, as well as knowing thestudents' self-concept of mathematics learning with the Open-ended approach. The method was used in this study is a quasiexperimentalmethod, with a pretest-posttest control group design research. The study was conducted in one of the junior highschools in Bandung Regency. The population in this study were all seventh grade students of one of the junior high schoolsin Bandung Regency, while the samples were students of class VII-A as an experimental class and class VII-D as a controlclass. The experimental class is treated with Open-ended approach and the control class is treated with Problem-BasedLearning model. The instrument were a mathematical problem solving ability test, student questionnaire and teacherobservation sheet. In this study quantitative and qualitative data were obtained. Based on the analysis of the results ofmathematical problem solving tests, the conclusion is the improvement of students' problem solving ability using the Openendedapproach is no better than the students who use the Problem-Based Learning model in mathematics learning. Althoughthe improvement of problem solving in experimental class students is not better than control class students, but at least theOpen-ended approach has a positive effect on students' self-concept.