Recently, the role of mitochondrial DNA (mtDNA) se quences in taxonomy and phylogenetic inference has become contentious, and two extreme viewpoints have emerged, although, as far as we know, never juxta posed prior to this article. One position criticizes the use of mtDNA because the marker suggests mislead ing patterns of variation; specifically, phytogenies that are inconsistent with those derived from nuclear gene sequences in the context of species relationships among closely related taxa (Ballard and Whitlock, 2004; Shaw, 2002). The other extreme, the DNA barcode77 move ment, espouses the sole use of small fragments of a single mtDNA gene, cytochrome c oxidase I (COI), to identify most of life (Hebert et al., 2003a). The intention of this article is to present the disadvantages of these two ex treme viewpoints and to argue for an integrated role for mtDNA, one that takes advantage of mtDNA's strengths but also accounts for its shortcomings by using it in con cert with other independent data sources (e.g., nuclear DNA, cytosystematic, morphological, behavioral). We are against the abolition of the use of mtDNA in phyloge netics but also against its narrow use in barcoding as cur rently defined. We demonstrate why neither viewpoint is particularly productive and emphasize how analysis of mtDNA can be an important tool in the context of both taxonomic and phylogenetic studies. The use of mtDNA in phylogenetics has been con tentious since it became clear that individual gene and species phylogenetic trees are not always congruent (Avise, 2004; Avise et al., 1983; Avise and Saunders, 1984), and discrepancies between nuclear and mtDNA inheri tance patterns have been well documented (see Funk and Omland, 2003). Recently, there has been a striking dis cord on the suitability of mtDNA in phylogenetics and taxonomy. Some (Hebert et al., 2003b) have argued that mtDNA barcodes, or approximately 600-bp segments of a small and discrete part of the genome, can be used to identify all of life, though the issue is actively debated (Hebert et al, 2003, 2004; Lipscomb et al, 2003; Moritz and Cicero, 2004; Pennisi, 2003; Scotland et al., 2003; Seberg et al., 2003; Will and Rubinoff, 2004). In sharp con trast to advocates of the use of mtDNA barcode segments alone for the identification of all biodiversity, others cau tio against use of certain mtDNA genes in some taxa (Lin and Danforth, 2004; Thalmann et al, 2004), whereas Ballard and Whitlock (2004), referencing studies such as that of Shaw (2002), question the utility of mtDNA for any study of systematics or phylogenetics. They ad vocate elimination of mtDNA from phylogenetic stud ies, sugg sting that it is risky to infer general patterns from an idiosyncratic small fraction of the genome (page 731). The implication is that mtDNA can be inaccurate or diverges from the true phylogeny We argue that most phylogenetic studies, other than the handful employ ing a large-scale, multigenomic approach (Rokas et al., 2003), may suffer drawbacks from using only a small fraction of the total genome. But it strikes us as mislead ing t str ngly criticize use of a character source that has yielded over 25 years of phylogenetic data, the majority of which have proven useful, with dozens of published oligonucleotide primer sets and approximately 250,000 entries available in GenBank, DDBJ, and EMBL, includ ing complete mtDNA genomes for more than 800 species (http://www.mitomap.org/euk_mitos.html). The opti mal role for mtDNA is somewhere between these ex