Metazoan mitochondrial genomes provide a model system for evolutionary genomic studies. Due to their relatively small size, sequences for complete mitochondrial DNAs (mtDNAs) are now available for animals over a large phylogenetic spectrum, including recent additions for the three classes of Porifera and various cnidarians. This wealth of data gives us a unique opportunity to infer the mechanisms underlying the evolution of genomes. In this dissertation, I will address the example of extensive parallel evolution between two distantly related groups, the Bilateria and Class Hexactinellida (Phylum Porifera; glass sponges), citing the specific examples of degenerated tRNA secondary structures and changes in the genetic code. Much of the plasticity of metazoan mtDNA is linked to the variation of tRNA genes, and I will hypothesize upon the mechanisms that bring about this variation. I will also discuss possible mechanisms involved in the mitochondrial genome reduction of Nematostella vectensis (Class Anthozoa, Phylum Cnidaria), where all but two tRNAs are lost from the mtDNA. Finally, mitochondrial genome data from the Hexactinellida indicates +1 translational frameshifting in protein coding genes, which has only rarely been reported in the mt-genomes of animals. The application of molecular phylogenies both resolves questionable phylogenetic relationships inside Hexactinellida, but also suggests a pattern of evolution for various unusual mt-genomic features.