Reversible thinking ability is an individual ability to do a cognitive process by reverse action, contributing to one of the student’s competencies in solving mathematical problems. Many students encounter difficulties in solving problems that require reversible thinking due to the limitedness of teachers’ proficiency in mastering this ability. Different studies have suggested various teaching approaches to improve this ability in teacher education; however, teaching with a metacognitive approach has not yet been addressed in the recent study. Therefore, this study aims to improve prospective teachers’ reversible thinking ability through metacognitive-approach teaching. A quasi-experimental method with a pre-test, intervention, and post-test design was used in this study. The participants were 118 prospective mathematics teachers at two universities in Bandung, Indonesia, divided into two groups: 58 were in the experimental group, and the rest were in the control group. The participant’s initial mathematical ability (IMA) in both groups was categorized into high, moderate, and low. Using t-test, Mann-Whitney test, and Kruskal-Wallis test, the result shows a meaningful difference in the improvement of reversible thinking ability between prospective teachers who received the metacognitive-approach teaching and those who did not. However, there is no significant interaction effect between the teaching approach and initial math ability on the improvement of reversible thinking ability. To conclude, the recent study’s findings revealed that the metacognitive approach effectively improves prospective teachers’ reversible thinking in all IMA levels. Thus, it needs to be considered one of the alternative teaching approaches in higher education, especially in teacher education.