Because soil microbial communities are often altered by anthropogenic disturbance, successful plant community restoration may require the restoration of beneficial soil microbes, such as arbuscular mycorrhizal (AM) fungi. Recent evidence suggests that later successional grassland species are more strongly affected by AM fungi relative to early successional plants and that late successional plants consistently benefit from some AM fungi but not other AM fungal species. Many of these late successional species are also often missing in restorations despite being heavily seeded. To assess the effects of AM fungal composition within grassland restorations, we inoculated plots with six different AM fungal community treatments including one of four different AM fungal species isolated from a prairie, a mixture of all four fungal species, and a non-inoculated control. AM fungi were introduced by planting 16 different inoculated nurse plants into replicated plots. We also seeded the restoration with a diverse, 54 species prairie seed mixture. We found that AM fungal inoculation drove plant community composition; plots inoculated with certain AM fungal treatments were dominated by desirable prairie plants, whereas plots inoculated with other AM fungal species and the non-inoculated control were dominated by non-desirable plants including weeds and exotic species. Specifically, we found that many early successional species established well regardless of AM fungal inoculation, whereas the establishment and growth of many late successional species was strongly dependent on the presence of specific AM fungal species. Many conservative late successional species did not occur without inoculation. Overall, total plant community richness, diversity, and Floristic Quality Index were all significantly improved with AM fungal inoculation, whereas we observed that non-desirable plant abundance was significantly greater in the non-inoculated plots. Synthesis and applications. Our results suggest that the lack of late successional establishment reported in many previous restorations may be due to ineffective arbuscular mycorrhizal fungal communities at these sites. We conclude that the reintroduction of arbuscular mycorrhizal fungi from reference prairie environments could improve restoration outcomes by promoting plant diversity and richness, especially for desirable later successional plant species, while simultaneously inhibiting less desirable weedy plants.