Detecting and quantifying low-level variants in mitochondrial DNA (mtDNA) by deep resequencing can lead to important insights into the biology of mtDNA in health and disease. Massively parallel ("next-generation") sequencing is an attractive tool owing to the great depth and breadth of coverage. However, there are several important challenges to be considered when using this method, in particular: the avoidance of false discovery due to the unintended amplification of nuclear pseudogenes and the approach to delineating signal from noise at very great depths of coverage. Here we present methods for whole mtDNA genome deep sequencing (Illumina MiSeq) and short amplicon deep sequencing (Roche 454 GS-FLX).