This essay reviews comparative studies of animal mitochondrial DNA (mtDNA), with emphasis on findings made and ideas developed at Berkeley. It argues that such studies are bringing together two previous paths of progress in evolutionary biology. One path is that of those who worked far above the species level and were concerned with genealogical trees, time scales and the accumulation of new mutations on surviving molecular lineages. The other path is that of those who worked at and below the species level and were concerned mainly with population structure, migration and the frequencies of alleles that existed in an ancestral population. This fusion of paths is made possible by the high rate at which mutations accumulate on mtDNA lineages and by this molecule's uniparental and apparently haploid mode of inheritance. These properties make mtDNA a superb tool for building trees and time scales relating molecular lineages at and below the species level. In addition, owing to its mode of inheritance, mtDNA is more sensitive to bottlenecks in population size and to population subdivision than are nuclear genes. Joint comparative studies of both mtDNA and nuclear DNA variability give us valuable insights into how effective population size has varied through time. Such studies also give insight into the conditions under which mtDNA from one species can colonize another species.