Recent evidence suggests that biotic soil communities may play a critical role in structuring plant communities. With little exception, ecologists currently lack theory that incorporates belowground organisms and processes into models of plant community structure. The soil community feedback framework (sensu Bever 2003) makes explicit predictions for species-specific population dynamics based on competitor soil feedbacks on performance and competitive interactions in sterile conditions. In a greenhouse experiment, I tested the predictions of this model and the relative importance of soil communities on 1) plant species’ performance in monoculture and 2) on outcomes of pairwise interspecific competition. I grew 4 old-field plant species (Aster novae-angliae, A. pilosus, Solidago candensis, S. gigantea) factorially with their own soil community, each competitor’s soil community, and the absence of soil community (sterile controls).  In general, the presence of a soil community had strong effects on the performance and competitive outcome of all species, indicating soil community dependency. Interestingly, only the two Aster species were sensitive to soil community identity, with generally negative feedbacks from competitors’ soil. The predictions of the soil community feedback model roughly correspond to results from the interspecific competition trials, and therefore may explain the relatively low abundances of Aster species in a community dominated by Solidago canadensis.