A fundamental assumption of models for the maintenance of genetic variation by environmental heterogeneity is that selection favours alternative alleles in different environments. It is not clear, however, whether such antagonistic pleiotropy is common. We mapped quantitative trait loci (QTLs) causing variation for reproductive performance in each of three environmental treatments among a set of 98 recombinant inbred (RI) lines derived from a cross between two D. melanogaster laboratory strains. The three treatments were standard medium at 25°C, ethanol-supplemented medium at 25°C, and standard medium at 18°C. The RI lines showed highly significant genotype–environment interaction for the fitness measure. Of six QTLs with significant effects on fitness in at least one of the environments, five had significantly different effects at the different temperatures. In each case, the QTL by temperature interaction arose because the QTL had stronger effects at one temperature than at the other. No evidence for QTLs with opposite fitness effects in different environments was found. These results, together with those of recent studies of crop plants, suggest that antagonistic pleiotropy is a relatively uncommon form of genotype–environment interaction for fitness, but additional studies of natural populations are needed to confirm this conclusion.