A strong correlation exists between small-scale spatial patterns and intraspecific plant interactions for Achnatherum splendens (Trin.) Nevski. K2 point pattern function, a spatial analysis method designed to avoid the effects of environmental heterogeneity, was applied to analyze the spatial patterns and intraspecific interactions of Achnatherum splendens on a small scale (within 0.5 m). Sixty-three A. splendens quadrats were established and studied in an arid community dominated by Elaeagnus angustifolia L. and A. splendens in the northwest China. The quadrats were established at three density levels in three microenvironmental types and with seven replicates of each. The different responses of A. splendens were compared based on the three spatial density patterns, low, medium, and high, and three microenvironmental types, subcanopy, transitional, and open areas. The soil physicochemical properties of electricity conductance, soil organic matter, and soil bulk density, were measured in the three microenvironments to quantify the environmental stresses. The results show soil physicochemical stress increased along the subcanopy to transitional area to open area gradient. The subcanopy area had relatively low environmental stress as evidenced by low soil electrical conductivity, high soil organic matter, and low soil bulk density. A. splendens was clumped in only six of the 21 subcanopy quadrats, while in the transitional and open areas where environmental stress was high A. splendens was clumped in a small majority of the quadrats (11/21). The high environmental stress areas were defined as areas having high soil electrical conductivity, low soil organic matter, and high soil bulk density. A. splendens tended to be clumped in areas with increasing environmental stress along the subcanopy to transitional area to open area gradient. However, the spatial responses of A. splendens to environmental stress differed at the three density levels. At the low-density level, A. splendens had a clumped distribution in most quadrats at all three environmental stress levels. The clumped distribution proportions of A. splendens quadrats were 4/7, 7/7, and 7/7 in subcanopy area, transitional area, and open area, respectively. In the medium density level, the frequency of A. splendens quadrats with clumped distribution increased with the environmental stress. In these medium density level areas the proportion of clumped distribution of A. splendens populations were 2/7, 3/7, and 4/7 in subcanopy area, transitional area, and open area, respectively. In the high-density level, A. splendens was distributed randomly in most quadrats (except one in a transitional area) in all three environmental stress levels. Since the spatial pattern of A. splendens showed a clear tendency to be clumped along the density gradient rather than along the soil physicochemical stress gradient, this might suggest the spatial pattern of A. splendens was more influenced by density than by the stresses caused by severe soil conditions. We concluded the spatial pattern of A. splendens on a small scale responded to the environmental stresses differently based on population density. As the result, population density should be considered when analyzing the variations of spatial patterns and the occurrence of positive interactions along the stress gradient.