Heterozygosity–fitness correlations use molecular measures of heterozygosity as proxy estimates of individual inbreeding coefficients (f) to examine relationships between inbreeding and fitness traits. Heterozygosity–fitness correlations partly depend on the assumption that individual heterozygosity and f are strongly and negatively correlated. Although theory predicts that this relationship will be strongest when mean f and variance in f are high, few studies of heterozygosity–fitness correlations include estimates of f based on pedigrees, which allow for more thorough examinations of the relationship between f, heterozygosity and fitness in nature. We examined relationships between pedigree-based estimates of f, multilocus heterozygosity (MLH) and the probability of survival to hatch in song sparrow nestmates. f and MLH were weakly, but significantly negatively correlated. Inbreeding coefficient predicted the probability of survival to hatch. In contrast, MLH did not predict the probability of survival to hatch nor did it account for residual variation in survival to hatch after statistically controlling for the effects of f. These results are consistent with the expectation that heterozygosity–f correlations will be weak when mean and variance in f are low. Our results also provide empirical support for recent simulation studies, which show that variation in MLH among siblings with equal f can be large and may obscure MLH–fitness relationships.