Group sizes of overwintering colonies of narrow-skulled voles (Microtus gregalis) may affect predation risk by mammals. We examined the effect of different social group sizes on the predation risk of narrow-skulled voles in the Xilinguole typical steppe, Inner Mongolia, West Ujimqin Banner, from September 15 to 30, 2005. Furthermore, we examined the effects of predation by the steppe polecat (Mustela eversmanni) on the group size of the narrow-skulled vole within the sampling area. The number of narrow-skulled vole colonies was used to estimate group sizes of voles. The number fecal droppings and holes excavated by steppe polecats within the habitat of each narrow-skulled vole colony were recorded as indicators of predation pressure. A total of 102 voles were observed within the sampling area. The nonparametric Mann-Whitney U test was used to detect differences in social group sizes between the colonies with and without signs of polecat activity. The results showed that the rank sum of the colonies not surrounded by polecat feces was 3273.5, whereas colonies surrounded by polecat feces was 1,979.5 (U=113.5, Adjusted Z Value=-6.387, Adjusted P < 0.001). The rank sum of colonies with holes dug by polecats was 556, whereas the rank sum of colonies without the polecat excavation was 4697 (U=41, Adjusted Z Value=-3.525, Adjusted P=0.0004). The results of the Mann-Whitney test suggested significant differences in predator preference between the two groups of colonies. In addition, the frequencies of feces and excavation trails rise with the increase of group sizes. This difference indicated that the steppe polecat significantly preferred narrow-skulled vole colonies with high group sizes. The polecats tended to spend longer periods around colonies with several burrow entrances. Some polecats even attempted to dig out such colonies, and therefore, several fecal droppings and excavation trails could be found around these colonies. This finding indicated that the colonies with larger social group sizes experienced higher predator risk.