This study aims to describe: (1) the effectiveness of problem-based learning of the numbered head together (PBL-NHT) type and that of jigsaw (PBL-Jigsaw) type, and (2) the more effective teaching between PBL-NHT and PBL-Jigsaw in terms of learning achievement, critical thinking skills, and mathematics dispositions of class X students. This study was a quasi-experiment using a pretest-posttest with the non-equivalent multiple-group design. The instruments used were a math achievement test, critical thinking test, and mathematics disposition questionnaire. The data were analyzed using the multivariate (MANOVA) and independent sample t test. The results of this study show both the models are effective in terms of critical thinking skills and mathematics dispositions, but they are not effective in their learning achievement. The PBL-Jigsaw is more effective than PBL-NHT in mathematics dispositions, but does not differ in terms of the critical thinking skills and learning achievement.