Mitochondrial genomes of different eucaryotes are not all alike. We have examined mitochondrial translation products in a number of yeasts (Candida krusei, Hansenula saturnus, Rhodotorula glutinis, Rhodotorula rubra, Torulopsis glabrata and Saccharomyces cerevisiae) and in Prototheca zopfii, a colorless alga, in order to determine whether certain proteins are invariably synthesized within mitochondria, how different these proteins are, and what additional proteins, if any, might be synthesized by diverse mitochondria. Using a variety of techniques and criteria, including immunological analysis and peptide mapping, we show that all the yeasts studied, and probably P. zopfii as well, make versions of the 3 large subunits of cytochrome c oxidase. Not all of these oxidase subunits are equally closely related to their counterparts in S. cerevisiae, however. Mitochondria of some of the yeasts studied do not make, or make only small amounts of, a counterpart to Varl, a major mitochondrially made protein in S. cerevisiae. Mitochondria of P. zopfii possibly do not make an apocytochrome b. T. glabrata, H. saturnus and the two Rhodotorula species each make one or more proteins whose relationship, if any, to mitochondrial translation products of S. cerevisiae is not apparent. These results provide new information about mitochondrial diversity. Whereas mitochondria of all the organisms that we have studied devote the major part of their synthetic effort to making the three large subunits of cytochrome c oxidase, and probably make certain other proteins in common, they do not all synthesize a completely identical set of proteins.