The achievement of high school graduates' competence is continuously improved by the government in Indonesia, one of which is by implementing the Minimum Competence Assessment. The main problem in its application is the limitations of similar problem models provided in classroom learning so that students are not accustomed to solving them. The purpose of this research is to produce mathematical reasoning questions using the Minimum Competence Assessment model for junior high school students that are valid, reliable, and have a potential impact. This research is development research with two stages of developing Tessmer's model, namely: (1) Preliminary and (2) formative evaluation stage, which consists of self-evaluation, prototyping (expert reviews and small group), and field tests. The trial subjects of this study were 25 students of class VIII Bengkulu state junior high school 7 in Bengkulu City. The results of the study obtained questions that met valid criteria with an Aiken index each of which was more than 0.5. The questions meet reliable criteria with a count of more than 0.6 and are in the high category. The questions have a potential impact with sufficient students' selection abilities on average and students giving a good response of more than 50.00%. The suggestion from the research is that the teacher encourages choosing a real context to present the problem in mathematics learning.