Abstract: The rapid development and increasing accessibility of eye-tracking technology have contributed to addressing the question of how to enhance the accuracy and efficiency of mathematical cognitive assessment. A review of 151 recent international studies on the application of eye-tracking technology in mathematics education over the past decade was conducted from three dimensions: commonly used eye movement metrics, application themes, and eye movement evidence. It was found that, at the methodological level, eye-tracking analysis is shifting from general analysis toward subject-specific focus, yet its integration with artificial intelligence remains insufficient. In the context of numerical cognition research, the internal mechanisms of numerical processing can be revealed by eye-tracking technology, addressing cognitive process issues that traditional behavioral data cannot explain. In the study of learning strategies and instructional interventions, eye-tracking technology holds dual value in both outcome evaluation and process revelation. In the contexts of special learner populations and teacher professional development, the visual processing characteristics of different groups can be identified by eye-tracking technology, though research from the teacher perspective remains relatively weak. Based on these findings, four implications and prospects are proposed for the application of eye-tracking technology in mathematics education research in China.