Computational thinking is an essential ability for students in the 21st century. Therefore, this study described students' mathematical computational thinking process in terms of self-efficacy of eighth graders, consisting of 32 students. The applied instruments were self-efficacy questionnaires, computational thinking ability tests, and interview guidelines. The results of the questionnaire were grouped into three categories, namely high, moderate, and low self-efficacy. A student was selected for each category as the respondent using the purposive sampling method. The data were analyzed with the Miles and Huberman model, starting from reduction, presentation, and conclusions. The computational thinking process is indicated by the stages of abstraction, representation, decomposition, algorithms, and evaluation. The results of the research analysis showed that the computational thinking process of students in the self-efficacy category was high and moderately fulfilling the five stages well and informatively. In addition, there needs to be an emphasis and increase in cognitive activity at the representation stage, especially in students with low self-efficacy. The representation stage helps students to get initial ideas for solving problems. This will have a further impact on the achievement of the algorithm stage in the computational thinking process. Based on these findings, the study is recommended to design five stages of the computational thinking process. Future research can develop the steps of the computational thinking process in solving contextual problems.