Abstract: Forest fire is the key disturbance factor in forest ecosystem succession, and its occurrence intensity and fire spread rate are deeply regulated by the surface fule load. Especially in the areas where the fule layer is continuously distributed, the load level directly determines the fire risk level and fire behavior characteristics. Therefore, constructing high-precision fule load prediction model has become the core link in improving the accuracy of forest fire risk forecasting. Based on domestic and international literatures on forest surface fule load, this paper reviewed the research methods of fule load, the application and verification of surface fule load prediction models, and the factors affecting the accuracy of fule load prediction models. At present, the research on the factors affecting the dynamic changes in surface fule load in China was still not perfect. The paper highlighted three major bottlenecks in current research: First, the coupling mechanisms of multiple factors was insufficiently analyzed, and some key driving factors were not fully considered in model construction; Second, the construction of basic databases was lagging, and the characteristic parameter systems of typical fuel types had not yet been perfected; Third, the regional scale adaptability had not been verified, and the promotion of models faced challenges of spatial heterogeneity. In order to solve the above issues, it is necessary to strengthen the construction of dynamic monitoring networks, multi-source data fusion analysis, spatial heterogeneity characterization, and multi-scale model integration research in the future, so as to promote the development of forest fire risk prediction towards refinement and intelligence direction.