This research was motivated by the low mathematical problem solving ability of students. The purpose of this study was to determine the improvement of better mathematical problem solving skills in students whose learning uses the Guided Discovery Learning (GDL) model with students whose learning uses conventional models, as well as to determine students' attitudes towards mathematics learning using Guided Discovery Learning (GDL). The method used in this study is a quasi-experimental method with a nonequivalent pretest-posttest control group design. Based on the results of normalized gain data processing using t-tests and a significance level of 5%, results were obtained to increase the mathematical problem solving ability of students whose learning using the Guided Discovery Learning (GDL) model was better than conventional models. Based on the processing of student attitude questionnaire data obtained, it can be concluded that student attitudes are positive towards mathematics learning with the Guided Discovery Learning (GDL) model.