Abstract: Soil moisture is the key factor in the formation and evolution of vegetation pattern. The spatial heterogeneity of soil moisture has an important significance in understanding the response mechanisms of plants in arid grasslands to environment. Meanwhile, at small scales, vegetation condition is the important driving factor to the spatial heterogeneity of soil moisture. By the methods of geo-statistics, this paper studied the spatial variability of surface soil (0-30 cm) moisture content and vegetation coverage as well as their relationship in the Melica przewalskyidominant patches in degraded grassland on the northern slope of Qilian Mountains, Northwest China. In the M. przewalskyidominant patches, both the surface soil moisture content and the vegetation coverage were in line with normal distribution. In different layers at 0-30 cm depth, soil moisture content had a high degree of spatial heterogeneity, and 80.93%-87.34% of the heterogeneity was caused by spatial autocorrelation. The spatial autocorrelation in vegetation coverage was found within the distance from 4.09-6.91 m, whereas the spatial heterogeneity of vegetation coverage caused by stochastic factors at <1 m scale accounted for 12.66%-19.07%. The spatial heterogeneity of soil moisture content in M. przewalskyidominant patches showed an obviously layered and patchy structure, and the high and low moisture values had a mosaic distribution. Through physiological integration, M. przewalskyi affected, and in definite extent, altered the spatial distribution of soil moisture at small scales, and thus, achieved the maximum utilization of soil water resource, and improved the adaptability and competitiveness of M. przewalskyi populations in arid habitats.