Livestock grazing is one of the primary causes of disturbances to grassland ecosystems. Ecosystems pay dramatic ecological costs in terms of their composition, structure, and function because of overgrazing. These costs included decreases in plant biomass, reductions in species richness and the number of high quality forage species, and increases in the numbers of toxic species and the percentage of bare ground, as well as changes to the physical and chemical properties of their soil. Livestock grazing has an effect not only on plants and soil, but also on small mammals. Many studies have suggested that grazing affects rodents in terms of their individual conditions, breeding, population dynamics, community structure, and species diversity. These effects could be further complicated by a background of climate change. However, because of the limits of traditional statistical methods, most studies only test individual hypotheses for grazing impacts on rodents, and so few studies test multi-hypothesis frameworks. A study was conducted from 2006 to 2011 on sites experiencing grazing exclusion,rotational-grazing, and over-grazing to investigate the effects of grazing on the population densities of dominant rodent species in Alashan, Inner Mongolia, China. Rodent species were identified and monitored using a live-trapping method.We trapped nine species including Dipus sagitta, Stylodipus andrewsi, Allactaga sibirica, Phodopus roborovskii, Cricetulus barabensis, Allocricetulus eversmanni, Meriones meridianus, M. unguiculatus, and Spermophilus alaschanicus. D. sagitta, A. sibirica, and M. meridianus were found to be dominant in their local rodent communities. In addition, climate data and soil compaction, as well as the coverage, height, density, and biomass of vegetation were measured in the study areas. In this study, data on dominant rodent species population densities were combined with habitat factors to perform a multi-hypothesis test using structural equation modeling. We expected to find which habitat factors had the greatest effects on the rodent population. The results indicated that climate conditions had direct negative effects on the population densities of D. sagitta and A. sibirica. The vegetation shield had negative effects on the population density of A. sibirica, but positive effects on that of M. meridianus. Increasing soil compaction could indirectly promote population density of A. sibirica, but directly increased the population density of M. meridianus. The vegetation shield had the greatest direct effect on dominant rodents, but soil compaction had the most indirect effects. These results suggest that the population dynamics of dominance in desert rodent species have been driven by many processes, and that climate has been a key factor in these processes. However, the population dynamics of dominance in rodent communities has been primarily regulated by changes in grazing due to changing vegetation cover.