SEEKING A MORE integrative understanding of the coupled human-environment system to tackle critical issues is currently a key concern for policymakers and the scientific community, especially in developing countries. This study developed an agent-based model (ABM) to simulate spatiotemporal changes of land use functions (LUFs) in the environmentally fragile region Guyuan in western China to better understand local issues. We employed both the Framework of Participatory Impact Assessment (FoPIA) and a questionnaire survey to combine human factors with environmental data. We concluded that: (1) Our ABM well represented spatiotemporal LUF changes in Guyuan; and (2) Land abandonment is a critical problem in Guyuan. Thus, the promotion of land cultivation and the improvement of cultivated land-use efficiency is urgently needed. Our model has great potential for land-use policy scenario simulations and could benefit policymakers in recognizing land-use issues and recommending land management policies.