Observations of rapid shifts in mitochondrial DNA (mtDNA) variants between generations prompted the creation of the bottleneck theory. A prevalent hypothesis is that a massive reduction in mtDNA content during early oogenesis leads to the bottleneck. To test this, we estimated the mtDNA copy number in single germline cells and in single somatic cells of early embryos in mice. Primordial germ cells (PGCs) show consistent, moderate mtDNA copy numbers across developmental stages, whereas primary oocytes demonstrate substantial mtDNA expansion during early oocyte maturation. Some somatic cells possess a very low mtDNA copy number. We also demonstrated that PGCs have more than 100 mitochondria per cell. We conclude that the mitochondrial bottleneck is not due to a drastic decline in mtDNA copy number in early oogenesis but rather to a small effective number of segregation units for mtDNA in mouse germ cells. These results provide new information for mtDNA segregation models and for understanding the recurrence risks for mtDNA diseases.