The purpose of this study is to find out what types of problems and scaffolding should be set in problem-based learning to alleviate students’ naive theories. This study employed a sample of three classes with a problem-based learning process using three variations of problems and scaffolding. The findings suggested that there were significant differences among the three groups of students. The highest reduction in students’ naive theories occurred in classes that used open problems equipped with discussion scaffolding. Based on the interpretation of the video results of student discussions and the average posttest class using open problems equipped with discussion scaffolding, students’ naive theories were mostly reduced. Problems in problem-based learning are open-ended, which encourage students to use their naive theories to solve problems. With the scaffolding that has been designed, students’ character grows as scientists. Based on the interview results, the class that used closed problems equipped with scaffolding tended to use their notes while in high school, which he was not clear about, except for doing exam questions. However, in a classroom that uses open problems equipped with discussion scaffolding and in classes that use open problems without discussion scaffolding, it is stated that students like to solve problems in their real lives. Thus, open problems in problem-based learning are the key to successful learning, whereas discussion scaffolding strengthens the implementation of these open problems.