Cultivating geospatial thinking is a key focus in geography education. As an eff ective instructional aid, the teaching eff ectiveness of Desktop Virtual Reality (DVR) is moderated by individual diff erences among learners, such as cognitive style. Therefore, empirical research on whether DVR is effective in enhancing high school students' geospatial thinking, how it works, and for whom it is more eff ective can contribute to the implementation of individualized teaching and the advancement of educational informatization. This study employed independent samples t-tests and structural equation modeling analysis. Results indicate that compared to traditional image-based instruction, DVR teaching is significantly more effective in improving students' geospatial thinking; DVR primarily infl uences students' geospatial thinking and learning outcomes through the pathway of "instructional media-cognitive load-spatial presence-geospatial thinking-learning outcome"; diff erent cognitive styles exert a diff erential moderating eff ect only on the "spatial presence-geospatial thinking" pathway. Based on the fi ndings, this study proposes personalized recommendations for teaching practice.