This study is a pre test-post-test experimental control group design having a goal to analyze the role of problembased learning on students' mathematical problem-solving ability (MPSA) and self-confidence (MSC). The study involves 66 tenth grade students, a mathematical problem-solving test, a mathematical self-confidence scale, and perception of problem-based learning (PBL) approach scale. The study found that on MPSA, its gain, and on MSC students getting treatment with PBL approach obtained better grade than that of students taught by conventional teaching. The other findings, there was the high association between MPSA and MSC, and student performed positive opinion toward PBL approach.