ABSTRACT: It is well established that somatic mutations are present in mitochondrial DNA (mtDNA) of cells and tissues expressing a variety of different disease phenotypes. The contribution of these changes to the disease phenotype is not at all clear, if there is a contribution at all. The primary obstacle to assessing the causal or consequential relationship between somatic mutations in mtDNA and disease is the retrospective nature of the analysis. Many of the reported somatic mutations in mtDNA are located in a region known as hypervariable region 3 (HVR3). Often there is concordance in position and substitution between single-nucleotide polymorphisms (SNPs) and somatic mutations observed in HVR3, leading to the possibility that these changes are not of functional significance. We have obtained biochemical evidence for a role of HVR3 in mitochondrial transcription. Our hypothesis now is that SNPs and somatic mutations in HVR3 may lead to misregulated transcription and contribute to disease phenotypes. To test this hypothesis, we have used the system described by Ridky et al. (Nat Med. 2010 Dec; 16(12):1450-5.) to establish squamous cell carcinomas ex vivo starting from primary human keratinocytes. This system permits a prospective analysis of changes in mtDNA sequence during the process of tumorigenesis. We observe mutations in HVR3 reminiscent of those observed retrospectively. To our knowledge, this is the first prospective analysis of the evolution of mtDNA during the transition from a normal to a disease phenotype. The implications of this system and our findings on the consequence of HVR3 mutations will be discussed.