This study aimed to investigate the effectiveness of: (1)the problem-based learning (PBL) method; (2)the direct teaching method; (3)the PBL method compared with the direct teaching method, for the low prior knowledge students; (4)the PBL method compared with the direct teaching method, for the high prior knowledge students. This research is a quasi experimental with the pretest-posttest non-equivalent control group design. The independent variabel was teaching method, the score on pre-test as a covariate, and the students. achievement as the dependent variable. The research population comprised year XI students of the Light Vehicle Engineering SMK Muhammadiyah Prambanan, totaling 137 students, divided into four classes. This study used a sample of two XI classes, totaling 64 students. The instrument was an achievement test in the multiple-choice form. The validation was conducted by analyzing the validity of the test items, difficulty level, discrimination index, and the functioning of the detractors. The result was that the 40 items used were all valid and the reliability was high (á=0.921). The data were analyzed using the analysis of covariance (ANCOVA). This study conclude that, in the competence of repairing the cooling system by considering the students' score on pre-test as prior knowledge: (1) on the whole, the PBL method is more effective than the direct teaching methods, (2) the students who have high prior knowledge are more effective to be taught using the PBL method than the direct teaching method, (3) the students who have low prior knowledge are not difference in effectiveness to be taught using the PBL method and the direct teaching method, and (4) there is no difference in effectiveness between low prior knowledge students and high prior knowledge students who were taught using the PBL method.