Abstract: Based on the forest resource inventory data in 2016, 18 vegetation subtypes in Yunnan Province were selected. The MaxEnt model with optimized parameters was used to predict the distribution of vegetation subtypes in combination with climate, topography, soil and other variables. The differences between the actual distribution and the theoretical distribution and their response mechanism to landscape disturbance were explored. The results showed that the model prediction accuracy(AUC) was above 0.9 and the TSS was above 0.6. Isothermality(bio3), the lowest temperature of the coldest month(bio6), annual precipitation(bio12) and altitude(Elev) were the main driving factors. The actual distribution was dominated by semi-humid evergreen broad-leaved forests, while the predicted distribution was dominated by warm temperate coniferous forests. This difference was mainly due to the interference of human activities with vegetation distribution, resulting in the areas suitable for the growth of warm temperate coniferous forests being actually covered by semi-humid evergreen broad-leaved forests. The analysis of the degree of landscape disturbance showed that the distribution of vegetation subtypes was mainly affected by low to moderate disturbances, and the ecosystem showed a certain degree of adaptability and resilience. In the future, it is necessary to accurately control the degree of disturbance and maintain the stability of the ecosystem to cope with the potential threats brought by climate change and human activities. This study provides a scientific basis for regional ecological protection and vegetation restoration, and has important guiding significance for optimizing resource management and consolidating the ecological security barrier function in Yunnan Province.