To accurately characterize the spatial and temporal heterogeneity of water yield in Qilian Mountains National Park and explore the influencing factors and driving mechanisms affecting water yield. The InVEST model was used to visualise the spatial and temporal pattern of water yield in the Qilian Mountains National Park from 1990 to 2018 based on the parameters of precipitation, evapotranspiration of reference crops, soil depth, root depth, and land use type, using a geo-detector and combined with a scenario simulation methods to assess and quantify the effects of climate, land use, vegetation, terrain, and other factors on the spatial heterogeneity and changes in water yield in the Qilian Mountains National Park. The results showed that: (1) From 1990 to 2018, the water yield in the study area increased significantly with a trend of 0.56×10⁸ m³/a, with a spatial increasing trend from west to east. (2) Farmland and Forestland had the largest water yield depth, and Unused land and Grassland were the main contributors to water yield in the study area. The high-value water yield depth areas were located at high altitudes between 3 200 m to 4 600 m. (3) Precipitation was the first driver of spatial heterogeneity of water yield, and the interaction with actual evapotranspiration and land use type dominates. (4) From 1990 to 2018, climate change was the main driver of changes in water yield, with a positive effect on total water yield, followed by land use change, which mainly affects the spatial distribution of water yield change but had little effect on total water yield. The research results can further provide the basic theoretical for the water resources management and sustainable development of Qilian Mountain National Park.