Oxygen, either molecular oxygen or a reduction adduct, can tightly bind in the vicinity of the two forms of trivalent nickel occurring in hydrogenase from Chromatium vinosum, as evident from studies with 17O-enriched O2. This oxygen is not in the first coordination sphere of nickel. As has been reported earlier for hydrogenase from Desulfovibrio gigas (Fernandez, V.M., Hatchikian, A.C., Patil, D.S. and Cammack, R. (1986) Biochim. Biophys. Acta 883, 145-154), also the relative activity of the C.vinosum enzyme correlates well with the presence of only one of the two Ni(III) forms in the oxidized preparation. These results make it less likely that a specific oxygenation of only one of the Ni(III) forms would be the reason for the reversible inactivation of nickel hydrogenases by oxygen. Reaction of H2-reduced enzyme with 13CO now demonstrated beyond doubt that: (i) One 13CO molecule is a direct ligand to nickel in axial position; and (ii) hydrogen binds at the same coordination site as CO. It can also be concluded that hydrogen is not bound as a hydride ion, but presumably as molecular hydrogen. A simple way to explain the EPR spectra from the 13CO-adduct of the enzyme is to assume a monovalent state for the nickel.