Plant communities are often structured by interactions among species, such as competition or facilitation. If competition is an important factor that controls the presence and absence of species within intact communities, then a competitive hierarchy, a ranked order from competitive dominant to competitive subordinate, should predict the composition of intact communities. We tested whether a competitive hierarchy derived from pairwise comparisons accurately predicts species abundances within a constructed polyculture community consisting of seven species common to old-field plant communities. We first conducted a pot experiment in field conditions wherein we grew the species in all possible combinations, then created a competitive hierarchy derived from both competitive effect and competitive response for each species. Concurrently, at the same site in native field soil, we constructed polycultures consisting of the same seven species and calculated an abundance hierarchy based on foliar cover, biomass, and an index of species performance. The competitive hierarchy was not concordant with the abundance hierarchy, indicating that simple pairwise comparisons may not account for other factors that influence the abundance of species within relatively complex communities.