The structure of grassland communities can vary widely in response to heterogeneous habitat variables. In this study we document plant communities, soil types, and site characteristics for 12 Palouse prairie remnants in southeastern Washington and northern Idaho. We used general linear models to test the predictive value of 6 biophysical variables (slope, aspect, distance from edge, edge type, vegetation structure, and soil type) on 3 plant community metrics: species richness, Simpson's diversity index, and the dominance of exotic species. From full models including all variables, we used Akaike's information criterion (AIC) to select the best model for each metric. Aspect and vegetation structure were significant predictors of species richness (R² = 0.08) and diversity (R² = 0.09), while aspect, soil type, distance from edge, and edge type (type of adjacent matrix habitat) influenced the dominance of exotic species (R² = 0.28). Additional soil sampling determined that the depth to a restrictive layer also had significant impacts on the dominance of exotic species (R² = 0.31). This study confirms that biophysical characteristics influence Palouse prairie plant communities and may help set research and conservation priorities for isolated and uninventoried remnants.