The recovery of native communities after cultivation may be constrained by (a) the failure of species to reach a site or (b) their failure to survive once there. Although seed addition is a common method to test for seed versus microsite limitation, most studies do not follow populations beyond seedling establishment, nor do they measure seed dispersal. We examined dispersal across native grassland/old field boundaries and investigated the relative importance of seed and microsite limitation across multiple life-history stages and generations. Seed trapping showed little movement of native seeds into old fields and that most species had extremely localized dispersal. Consequently, there was no pattern of seed density with distance from boundaries, and similarity between the seed rain and standing vegetation was moderate to high. Seed addition showed that two annual species were able to establish in all, and flower in most, subplots in the first year, and that seedling establishment increased with sowing density, consistent with seed limitation. However, the relative importance of microsite limitation increased over the lifespans of the species. Density dependence reduced the number of flowering plants, resulting in a large decline in seedling density in the following generation. This decline continued so that the initial positive effect of sowing density on seedling numbers disappeared by the fourth generation and hence the persistence of populations is uncertain. Thus, by monitoring seed dispersal and following experimental populations beyond seedling establishment, we showed that dispersal limits species distributions, but microsite plays an important role in limiting population growth and persistence.