The genetic compatibility hypothesis proposes that females should mate with genetically dissimilar males whose alleles best complement their own, resulting in greater offspring heterozygosity. It predicts that genetic similarity between social pairs will be positively related to the proportion of extra-pair young within broods and negatively related to hatching success. We tested these two predictions in tree swallow (Tachycineta bicolor (Vieillot, 1808)) pairs (n = 72). Tree swallows have one of the highest rates of extra-pair paternity among socially monogamous passerines. Contrary to expectation, genetic similarity of a social pair, as measured by the band-sharing coefficient (estimated from multilocus DNA fingerprints), tended to be negatively related to the proportion of extra-pair young within broods, but failed to predict hatching success. When including only the subset of nests for which we had complete genotyping data (n = 37), we again found a significant negative relationship between genetic similarity and the proportion of extra-pair young within broods. Genetic similarity did not differ significantly between nests with and without extra-pair young, nor did it differ between nests with total versus partial hatching success. Overall, our data do not support the genetic compatibility hypothesis in tree swallows, and in fact show some evidence against it.