Wagner et al. have recently introduced much-needed data to the debate on how complexity of the genotype-phenotype map affects the distribution of mutational effects. They used quantitative trait loci (QTLs) mapping analysis of 70 skeletal characters in mice and regressed the total QTL effect on the number of traits affected (level of pleiotropy). From their results they suggest that mutations with higher pleiotropy have a larger effect, on average, on each of the affected traits-a surprising finding that contradicts previous models. We argue that the possibility of some QTL regions containing multiple mutations, which was not considered by the authors, introduces a bias that can explain the discrepancy between one of the previously suggested models and the new data.