Mitochondrial DNA is passed from mother to child, so it is uniparental genetic marker that traces matrilines all of the way back to a hypothetical mitochondrial eve is the most recent shared matrilineal ancestor of every human now living.
Most mtDNA, and indeed most DNA generally, is identical for all humans, so describing a person's mtDNA by reporting a person's entire genome including the parts shared by all humans would be an exceedingly inefficient approach.
Instead, what geneticists do is establish a common reference genome and report the way some other mtDNA sequence differs from the reference genome. Until now, the standard reference genome, called the Cambridge Reference Sequence (CRS for short), was the mtDNA of some random white European person who probably went to Harvard, at one extreme of a particular part of the mtDNA phylogeny. When the CRS was adopted as the international standard in 1981, of course, we knew a great deal less about the mtDNA sequences that existed in the world and how they were related to each other.
The new reference sequence is instead a hypothetical one, constructed by creating an mtDNA tree including all known human mtDNA haplogroups, and bit by bit working back towards their common ancestor. The last few steps involve mtDNA sequences that have never been seen in anyone, by can be logically reconstructed as necessary to bridge the differences between the most basal branches of the mtDNA tree that is observed. In other words, they have reconstructed the mtDNA genome of mitchondrial Eve who would have lived something on the order of 177,000 years ago.
Essentially everyone in the world alive today has between ten and fifty mutations that distinguish them from mitochondrial Eve and the new method for scientifically describing mtDNA sequences in the most rudimentary form will be to list those ten to fifty mutations from the mtDNA Eve reference sequence called the Reconstructed Sapiens Reference Sequence (RSRS for short). All of the mtDNA sequences that are not predominantly African or of relatively recent African origin (i.e. mtDNA haplogroups the have left Africa only within the last 10,000 years or so), have at least thirty mutations from the RSRS. The combination that all non-African mtDNA types have in common (L3), which can function as a reference sequence for non-African mtDNA sequences, is estimated at about 67,000 years old based upon its thirty or so mutations from the RSRS.
Mutation rate dating of the emergence of mtDNA haplogroups is more art than science, but those estimates aren't wildly out of line with archaeologically based dates for an out of Africa event (modern scholarship tends to point to a bit earlier dates, but it is possible that some of the earliest Out of Africans don't have matriline ancestors in modern populations). The average mutation rate for mtDNA is on the order of one mutation per about 3,500 years. This isn't the post where I'm going to obsess about the calibration or miscalibration or even inherent limitations of mutation rate mtDNA dates and these dates in this post are "for entertainment purposes only."
While the exact dates at which particular branches of the mtDNA tree diverged from each other are debatable, the number of mutations that exist from the RSPS in any given mtDNA sequence are not, and there is very little room for much adjustment to the mtDNA phylogeny from which the RSPS was deduced. The fact that neither of these facts are subject to much reasonable dispute is a remarkable accomplishment of science generally and genetics in particular.
UPDATE: A more detailed post at Ethio Helix makes clear that some mtDNA haplogroups have far more than fifty mutations.