Abstract: This study was mainly done on two dominant species, Stipa krylovii and Leymus chinensis, and on community of Xilinguole typical steppe in Inner Mongolia. The field experiment was carried on two kinds of plots, which while ones of them were enclosed and not grazed for many years, and the others were kept on grazing. With setting treatment of experiment on three water gradients, the growth characteristics of grass under the appropriate water conditions, effects of water stress, and impact of grazing and other environmental factors were analyzed in the study. The results of research showed that: (1) Under appropriate water conditions, height and biomass maximum of Stipa krylovii and Leymus chinensis were 87.9 cm, 40.3 cm , 280g/㎡and 54.2 g/㎡, respectively; the largest biomass, cover-degree and LAI of community were 467.6g/㎡, 91% and 2.7. (2) height of a species, aboveground biomass of Stipa krylovii and community were in line with Logistic growth model, while the growing curve of aboveground biomass of Leymus chinensis not manifested obvious S-type characteristics. And binomial and trinomial fitting were good in fitting analysis for cover-degree and LAI of community. (3) Water stress produced obvious negative effects on the grassland vegetation, and it showed worse gradually with increase of drought severity; the biomass response to water stress was stronger than to height, and water sensitivity of Leymus chinensis was greater than of Stipa krylovii ; with the soil relative humidity dropping on 40-50%, the largest biomass and LAI of community reduced nearly 50%, and 30% reduced on community coverage. When water was sufficient, 2500-2700 ℃was the accumulated temperature(>0℃) for best growth of Stipa krylovii. (4) Except for aboveground biomass of Leymus chinensis, the growth characteristics of communities and species on grazing plots were better than the enclosed plots. With regard to biomass and stem number of Leymus chinensis on per unit area in grazing plots, they were respectively less 25% and 66.8% than in the enclosed plots.