As a typical ecological functional zone in northwest China, the Qilian Mountains play an important role in soil and water conservation research for alpine and semi-arid areas. Soil anti-scourability (AS) is an important indicator of the ability of soil to resist flow erosion. To investigate the AS and the factors that impact in typical grassland ecosystems on the Qilian Mountains, the sediment yield of surface runoff was set as an index, artificial field flume experiments were carried out, and correlation, regression analysis, and path analysis were used to analyze the effects of topography, vegetation, and soil properties on AS. Correlation analysis showed that the AS was positively correlated with altitude, vegetation coverage, root density, and sand content at P < 0.01, negatively correlated with slope and silt content at P < 0.01, and positively correlated with vegetation diversity and above-ground biomass at P < 0.05. However, it had no significant correlation with average height of grass. Correlation analysis also showed that there were correlations between some factors (e.g., vegetation coverage and above-ground biomass, altitude, and vegetation coverage), which suggested that there was multicollinearity among factors. To eliminate the multicollinearity and identify the crucial factors, path analysis was used. Path analysis showed that vegetation coverage and root density were crucial factors for the AS of grass ecosystems on the Qilian Mountains. The direct and indirect path coefficients of vegetation coverage were -0.660 and -0.075 respectively, and the direct and indirect path coefficients of root density were -0.286 and -0.174 respectively. This indicates that the influence of vegetation coverage on AS was mainly an intense direct effect, whereas the influence of root density had a weaker direct effect, and there were quite a few non-negligible influences with indirect effects on the AS of root density. The overall contributions of vegetation coverage and root density to the regression equation's R² were 0.485 and 0.132, respectively. This suggests that vegetation coverage was a more crucial influence factor than root density. Regression analysis was used to determine the relationships between AS and vegetation coverage, and AS and root density. The AS increased obviously with vegetation coverage and root density, the exponential function y=19.73e^-0.02x-3.12(R²=0.57,P < 0.001)could be used to express the relationship between the sediment yield of surface runoff and vegetation coverage, and the exponential function y=11.15e^-191.57x+0.8(R²=0.69,P < 0.001)could be used to express the relationship between the sediment yield of surface runoff and root density. This research indicated that vegetation coverage and root density should be considered key factors in the search for soil erosion rules and benefit evaluations on the Qilian Mountains. Thus, it could be used to increase vegetation coverage and root density of grassland ecosystems by protecting existing vegetation and spreading seeds of plants with well-developed root systems to control soil erosion on the Qilian Mountains. It also provided valuable information for general ecological research.