Plant invaders have been suggested to change soil microbial communities and biogeochemical cycling in ways that can feedback to benefit themselves. In this paper, we ask when do these feedbacks influence the spread of exotic plants. Because answering this question is empirically challenging, we show how ecological theory on 'pushed' and 'pulled' invasions can be used to examine the problem. We incorporate soil feedbacks into annual plant invasion models, derive the conditions under which such feedbacks affect spread, and support our approach with simulations. We show that in homogeneous landscapes, strong positive feedbacks can influence spreading velocity for annual invaders, but that empirically documented feedbacks are not strong enough to do so. Moreover, to influence spread, invaders must modify the soil environment over a spatial scale larger than is biologically realistic. Though unimportant for annual invader spread in our models, feedbacks do affect invader density and potential impact. We discuss how future research might consider the way landscape structure, dispersal patterns, and the time scales over which plant-soil feedbacks develop regulate the effects of such feedbacks on invader spread.