Traditional rodent monitoring is time-consuming and labor-intensive, and the accuracy of the results is not high. In this study, the maximum entropy (MaxEnt) method is used to model the spatial distribution of rodents in a typical grassland (Jarud Banner) of Inner Mongolia, China, and the spatiotemporal distribution pattern and driving mechanism of the occurrence probability of rodent damage in Jarud Banner during 2016–2021 is analyzed. The results show that: (1) A spatial distribution model constructed for grassland rodent damage at the community scale accurately predicts the potential spatial distribution of rodents. (2) The degree of the relationship between rodent damage and environmental factors is not constant but shows a certain variation pattern. (3) Spatially, the occurrence probability of rodent damage in Jarud Banner shows a pattern of low in its central part and high in its northern and southern parts. (4) Temporally, the occurrence probability of rodent damage in Jarud Banner shows a trend of decreasing in its northern and central parts and slightly increasing in its southern part, with the occurrence probability increasing in 2019 and 2021. The results provide new information for studying the spatial distribution of rodent damage and offer important reference for rodent pest control efforts.