Recent findings on feedback between plants and soil microbial communities have improved our understanding of mechanisms underlying the success and consequences of invasions. However, additional studies to test for feedback in the presence and absence of interspecific competition, which may alter the strength or direction of feedbacks, are needed. We tested for soil microbial feedback in communities of the invasive grass Microstegium vimineum and commonly co-occurring native plant species. To incorporate competitive context, we used a factorial design with three plant treatments (M. vimineum alone, M. vimineum with the native plant community, and the native community without M. vimineum) and two soil inoculum treatments (experimentally invaded and uninvaded soil). When competing with M. vimineum, native communities were 27% more productive in invaded than uninvaded soil. In contrast, soil type did not significantly affect M. vimineum biomass or fecundity. At the community level, these results indicate a net negative soil microbial feedback when native plants and M. vimineum are grown in competitive mixture, but not when they are grown separately. Since positive, not negative, feedback is associated with dominance and invasion, our findings do not support plant-soil feedback as a driver of invasion in this species. Our results do show that the importance of soil feedback can change with competitive context. Such context-dependency implies that soil feedback may change when competitive interactions between natives and invading species shift as invasions progress.