The handicap hypothesis proposes that male signals provide reliable information to females because only males of high condition provide high-quality mating benefits and can afford the costs of producing attractive signals. In the context of direct benefits, the handicap hypothesis predicts that benefit quality and signal attractiveness will positively covary among genotypes, positively covary among environments, or be affected by congruent genotype–environment interactions. The latter should occur if the relative condition of a genotype is environment-dependent. We tested these predictions in the variable field cricket, Gryllus lineaticeps. An interaction between male family and nutritional environment affected the expression of a costly signal preferred by females, while only male family affected direct benefit quality. These noncongruent effects of family and nutritional environment are inconsistent with the handicap hypothesis, and appear to have resulted from variation among nutritional environments in the relationship between signal attractiveness and benefit quality. Surprisingly, signal attractiveness was positively correlated with benefit quality when males experienced a low nutrition environment but negatively correlated with benefit quality when males experienced a high nutrition environment. As a result, female choice for direct benefits may be difficult, particularly in heterogeneous environments, unless females can assess the environmental histories of males.