A long-standing hypothesis predicts that closely related individuals are likely to compete with each other more strongly than are distantly related individuals. This hypothesis rests on the largely untested assumption that phenotypic similarity increases with relatedness, leading to a greater overlap in resource use, and, potentially, reduced local co-existence. Competition should thus be more intense within than across related genera, between conspecifics than between congeners, and between siblings than between individuals from different maternal lines. We tested this hypothesis hierarchically in a pairwise competition experiment with two genera of old-field herbaceous perennials, Solidago and Symphyotrichum (Asteraceae), each represented by three co-occurring species. For all species, an individual was either grown singly, with a sibling, with a conspecific from a different maternal line, with one of the other congeners, or with a species from the other genus. After 11 weeks, we calculated relative competition intensity using above- and below-ground biomass. Contrary to predictions, competition was not consistently more intense between conspecifics than congeners, nor between congeners than across the two genera. In addition, across all six species, sibling plants did not consistently compete more strongly than plants from different maternal lines. Similarly, root-shoot allocation patterns were species-specific rather than following the hierarchy of relatedness. These results suggest that relatedness may not reliably predict competition intensity, even within a genus or species. Finally, we compared these results with co-occurrence patterns for the six species in nine undisturbed old fields in central NY. At small spatial scales (0.25 m²), the observed frequencies of species co-occurrence differed from those predicted by their competitive rankings, suggesting that competition alone is insufficient to account for the observed patterns of co-existence.