Since the discovery of pathogenic mitochondrial DNA (mtDNA) mutations in the 1980's, these mutations have been shown to be involved in a number of human diseases and have caught increasing attention. The emergence of next generation sequencing technology has allowed us to do more comprehensive studies of these mutations in the mitochondrial genome, especially for low frequency heteroplasmic mutations. In my research projects, I combined experimental and computational approaches to investigate mtDNA heteroplasmies in different medical disciplines. I first conducted a literature review to summarize recent findings on the implications of mtDNA mutations in human diseases, with a focus on complex diseases such as cancer, neurodegenerative diseases and aging. My first project was to investigate mtDNA heteroplasmy and copy number variations in a general healthy population. This population-based study indicated that both mtDNA quality and quantity decreased with age. I further found that mtDNA copy number was associated with serum bicarbonate level and white blood cell counts, while aggregate heteroplasmy load was associated with blood apolipoprotein B level. These results suggested that maintaining optimal mtDNA copy number and preventing the expansion of heteroplasmy could promote healthy aging. In my second project, I first identified heteroplasmies in 466 pairs of DNA and RNA sequencing data. I verified that most of the heteroplasmies were transcribed to RNA regardless of their pathogenic potentials. I then experimentally tested whether the heteroplasmy frequencies could change over time. My test showed that a heteroplasmy with ~50% frequency could decrease to ~1% in only 28 days. Moreover, these observed heteroplasmy dynamics could significantly affect certain gene expression levels. My third project focused on mtDNA fragments circulating in blood stream in a cell-free format. I developed an experiment protocol tailored for sequencing short DNA fragments from plasma samples. After analyzing the sequencing data, I found that the fragment length of mtDNA in plasma is much shorter than that of nuclear DNA. We also demonstrated that mtDNA heteroplasmy was detectable in the plasma sample, suggesting its potential to serve as a biomarker in different clinical applications.