Scientific evaluation of ecological quality of desert steppe in North China is helpful for ecological restoration and desertification control.Taking Siziwang Banner as an example, this paper adopts the landscape ecological perspective, constructs the ecological quality evaluation system, and combines the comprehensive index method to research grassland ecological quality. In this paper, grassland ecological quality of desert steppe in 130 administrative villages is comprehensively evaluated and classified with the natural breaking method. After comparing the ordinary least square regression model, the effect of exogenous pressure on grassland ecological quality was evaluated by introducing the geographical weighted regression model. The main results were as follows:(1) The grassland ecological quality indexes ranged from 0.003 to 0.765 in the village area of Siziwang Banner, showing the regional difference in north high to south low in 2016. The spatial characteristics of High-High agglomeration, Low-Low agglomeration and Low-High agglomeration reflect the disequilibrium characteristics and spatial dependence.(2) The GWR model has the advantage of explaining spatial heterogeneity and processing data nonstationarity of grassland ecological quality. Compared with the estimated parameters, the prediction accuracy of the GWR model is higher than the OLS model. (3) The regression coefficient of six variables on grassland ecological quality showed that the effect of economic pressure and environmental pressure was obviously higher than that of population pressure and resource pressure.(4)The positive correlation factors of the change of grassland ecological quality are as follows:ecological land ratio, per capital amount of farmland and grassland grazing capacity. The distance from the nearest town, labor to force ratio and per capital GDP showed a negative correlation in turn. The results can be used as reference to grassland ecological protection and management.