This study aims to describe the thinking process profiles of high school students in solving numeracy problems from the perspective of self-efficacy. The study employed a qualitative, descriptive approach. The research subjects were six 11th-grade students at MA Ibadurrochman in Malang, selected through purposive sampling based on three categories of self-efficacy (high, moderate, and low). The instruments used included a self-efficacy questionnaire, a numeracy test equivalent to the Minimum Competency Assessment (AKM), and a semi-structured interview guide. Data analysis utilized the interactive model proposed by Miles, Huberman, and Saldaña. The results showed that students with high self-efficacy were able to fully complete all four stages of Polya’s problem-solving model: understanding the problem, planning a strategy, executing the plan, and checking the solution. Students with moderate self-efficacy were only able to master the understanding and planning stages and did not conduct a thorough review of their work. Students with low self-efficacy experienced significant obstacles starting from the problem-understanding stage, made procedural errors, and did not verify their answers at all. The conclusion of this study confirms that self-efficacy plays a crucial role in determining the quality of students’ thinking processes when solving numeracy problems.