Availability and heterogeneity of resources have a strong influence on plant community structure in undisturbed systems, as well as those recovering from disturbance. Less is known about the role of resource availability and heterogeneity in restored communities, although restoration provides a valuable opportunity to test our understanding of factors that influence plant community assembly. We altered soil nitrogen (N) availability and soil depth during a prairie restoration to determine if the availability and/or heterogeneity of soil resources influenced plant community composition in restored grassland communities. Plant community responses to three levels of N availability (ambient, enriched by fertilization, and reduced by carbon amendment) and two levels of soil depth (deep and shallow) were evaluated. In addition, we evaluated plant community responses to four whole plot heterogeneity treatments created from the six possible combinations of soil N availability and soil depth. The soil depth treatment had little influence on community structure during the first 3 years of restoration. Total diversity and richness declined over time under annual N enrichment, whereas diversity was maintained and richness increased over time in soil with reduced N availability. Non-native species establishment was lowest in reduced-N soil in the initial year, but their presence was negligible in all of the soil N treatments by the second year of restoration. Panicum virgatum, a native perennial C(4) grass, was the dominant species in all soil N treatments by year three, but the magnitude of its dominance was lowest in the reduced-N soil and highest in enriched-N soil. Consequently, the relative cover of P. virgatum was strongly correlated with community dominance and inversely related to diversity. The differential growth response of P. virgatum to soil N availability led to a higher degree of community similarity to native prairie in the reduced-N treatment than in the enriched-N treatment. There were no differences in plant community structure among the four whole plot-level heterogeneity treatments, which all exhibited the same degree of similarity to native prairie. Diversity and community heterogeneity in the whole-plot treatments appeared to be regulated by the dominant species' effect on light availability, rather than soil N heterogeneity per se. Our results indicate that a strong differential response of a dominant species to resource availability in a restored community can regulate community structure, diversity, and similarity to the native (or target) community, but the importance of resource heterogeneity in restoring diversity may be dampened in systems where a dominant species can successfully establish across a range of resource availability.