Abstract The two principal theories of the causal mechanism for inbreeding depression are the partial dominance hypothesis and the overdominance hypothesis. According to the first hypothesis, inbreeding increases the frequency of homozygous combinations of deleterious recessive alleles thereby decreasing fitness, whereas the overdominance hypothesis posits that inbreeding increases homozygosity and thus reduces the frequency of the superior heterozygotes. These two hypotheses make different predictions on the effect of crossing inbred lines: the overdominance hypothesis predicts that trait means will be restored to the outbred means, whereas the partial dominance hypothesis predicts that trait means will exceed those of the outbred population. I tested these predictions using seven inbred lines of the sand cricket, Gryllus firmus. Fourteen generations of brother-sister mating resulted in an inbreeding depression of 20–34% in four traits: nymphal weights at ages 14 days, 21 days, 28 days, and early fecundity. An incomplete diallel cross of these lines showed genetic variation among lines and an increase in all trait means above the outbred means, with three being significantly higher. These results provide support for the partial dominance hypothesis and are inconsistent with the overdominance hypothesis.