The purpose of this study was to determine whether students' mathematical problem-solving abilities by applying the Contextual Teaching and Learning (CTL) learning model were better than students' mathematical problem-solving abilities by applying conventional learning models. This research was motivated by the low mathematical problem-solving ability of students. This type of research was an experiment using a quantitative approach with a Noneequivalent posttest-only Control Group Design. The population of this study was all students of class XI Senior high school Negeri 1 Padang Gelugur. The technique used in this research was random sampling, so the sample of this research was class XI natural sciences major 1 as the experimental class and class XI natural sciences major 2 as the control class. The research instrument used was Post-test. The form of the test used was an essay with reliability of 0.399. Furthermore, the hypothesis was tested by a t-test. The results of hypothesis testing show that tcount > ttable was 7.17 > 1.67 so hypothesis one was accepted. Thus, it can be concluded that students' mathematical problem-solving abilities by implementing the CTL learning model were better than students' mathematical problem-solving abilities by applying conventional learning.