The most common type of genetic relationship between cytosolic and mitochondrial isoenzymes will probably be found to be divergent evolution from a common ancestral form. This is firmly established for the aspartate aminotransferases and less directly so in other cases. The two isoenzymes of aspartate aminotransferase have evolved at roughly equal rates at the level of total amino acid sequence but certain limited surface regions of the mitochondrial form have been much more highly conserved than corresponding regions in the cytosolic protein; these regions probably play a role in topogenesis of the mitochondrial isoenzyme. It is of interest that nearly all mitochondrial proteins are initially synthesised as precursors of molecular weight greater than the mature forms. In the case of aspartate aminotransferase, and possibly of other such isoenzymes, the N-terminus of the mature protein is nearly coincident with that of the cytosolic isoenzyme. Hence during evolution either the gene for the mitochondrial isoenzyme has gained an extra coding region for this N-terminal extension or, less likely, the structural gene for the cytosolic form has suffered a sizeable terminal deletion. Cytosolic and mitochondrial superoxide dismutases have not shared a common ancestral form as shown by the fact that their primary structures are completely unrelated. On the other hand, the mitochondrial and prokaryotic enzymes are clearly related. There is now, however, evidence to suggest that some prokaryotes possess a copper/zinc enzyme related to the eukaryotic cytosolic form. Hence the possibility arises that primitive prokaryotes possessed both proteins. The copper/zinc superoxide dismutase has been retained in the cytosol of eukaryotic cells and a few bacterial species.(ABSTRACT TRUNCATED AT 250 WORDS)