Windbreak and sand-fixing service function (WSSF) is an important protection function provided by ecosystems in arid and semi-arid regions. It is of great significance to explore its spatial and temporal evolution and its driving mechanism for regional ecological governance. However, there is still a lack of accurate understanding of the contribution and interaction of natural and human factors in the process of its spatial pattern formation and evolution. In this study, we chose Xilingol League as the study area and RWEQ (Revised Wind Erosion Equation) model was used to evaluate the spatial and temporal changes of its WSSF between 2000 and 2015. The contribution and interaction of natural and human factors to the formation and evolution of the spatial pattern of the WSSF were analyzed by using Geodetector. The results illustrated that the spatial distribution of the WSSF was significantly different, which showed that the WSSF was high in the temperate deciduous broad-leaved forest and shrub areas in the southeast, the WSSF was relatively high in the grassland meadow areas in the northeast and the WSSF was low in the desert areas in the west. The transfer of the WSSF was mainly from the low functional area to the lower middle functional area. On the whole, the amount of windbreak and sand-fixing showed an increasing trend, with the increase area of 115104 km² and decrease area of 8475 km². Soil type played the most important role in the spatial distribution pattern of the sand-stabilization service function, its P_D,SF value was the highest and had significant difference with other factors. Vegetation type, afforestation area, number of livestock, annual wind speed also had strong influence. The dynamically geographic detection results in the area of the function changes between 2000 and 2015 showed that both soil type and vegetation type had high explanatory power and were relatively stable, and the P_D,SF values of climate factors such as annual wind speed and precipitation in the enhanced function area showed a decreasing trend, the trend of human factors increased first and then decreased, while the variation trend of the weakened function area was opposite basically, and the interaction between the detection factors was significant.