Inbreeding load, a key parameter in evolutionary ecology, is frequently estimated by regressing fitness (or related traits) on inbreeding coefficient across population members. This approach assumes that inbreeding occurs randomly with respect to an individual's intrinsic ability to produce fit offspring; estimated loads might otherwise be biased by covariation between inbreeding and individual quality. This assumption, however, is rarely validated. We tested whether, in free-living song sparrows Melospiza melodia, an individual's observed kinship with its social mate (and hence the degree of inbreeding in which an individual participated) was correlated with specific phenotypic traits that are likely to indicate individual quality. Males (and to some extent females) that hatched earlier within their cohort, had shorter tarsi, bred earlier during their first year, or survived fewer years paired with more closely related mates and therefore produced relatively inbred offspring. These correlations arose because males with specific phenotypes were more closely related to the female population (and therefore more likely to pair with closer relatives under random pairing), and because males with specific phenotypes paired with closer relatives than expected. Such correlations could bias estimated inbreeding loads, and should be considered in quantitative genetic analyses of phenotypic variance in populations in which inbreeding occurs.