Several studies of tumors have revealed substantial numbers of clonally expanded somatic mutations in mitochondrial DNA (mtDNA), not observed in adjacent intact tissues. These findings were interpreted as indicating the involvement of mtDNA mutations in tumorigenesis. Such comparisons, however, ignore an important confounding factor: the monoclonal origin of tumors as opposed to the highly polyclonal nature of normal tissues. Analysis of recently published data on the incidence of somatic mutations in nontumor monoclonal cells suggests that, contrary to the prevailing view, the process of tumorigenesis may be accompanied by active selection against detrimental mtDNA mutations.