The complete amino-acid sequence of the copper-zinc superoxide dismutase of the Photobacterium leiognathi was determined. The fragmentation strategy employed included cyanogen bromide cleavage at its methionine residues and the only tryptophan residue. The S-carboxymethylated chain was further cleaved by means of trypsin, in order to obtain overlapping fragments. For sequence determination automated solid or liquid-phase techniques of Edman degradation were used. C-Terminal amino acids of the entire chain were determined after treatment with carboxypeptidase A. Comparison of the primary structure of this bacterial Cu-Zn superoxide dismutase with the established amino-acid sequences of the other eukaryotic Cu-Zn superoxide dismutases revealed clear homologies. Correspondingly, the Cu-Zn-binding amino-acid residues of the active centre were localized: His45, His47, His70, His79, His125 and Asp91. The two cysteine residues in position 52 and 147 were homologous to the cysteine residues, modelling the essential intrachain disulfide bridge of the corresponding bovine enzyme. As only 25-30% of aligned sequence positions were found to be identical, the enzyme of P. leiognathi shows only a remote phylogenetic relationship towards eukaryotic Cu-Zn superoxide dismutases. When compared to the established phylogenetic tree of the cytochrome c family, this indicates a separate evolution of Cu-Zn superoxide dismutase in Photobacterium. Therefore, a natural gene transfer from the eukaryotic host (ponyfish) to the prokaryotic photobacterium, which Martin and Fridovich postulated 1981 (J. Biol. Chem. 256, 6080-6089) on the basis of amino-acid compositions, can be excluded.