The molybdenum iron-sulphur protein originally isolated from Desulfovibrio gigas by Moura, Xavier, Bruschi, Le Gall, Hall & Cammack [(1976) Biochem. Biophys. Res. Commun. 72, 782-789] has been further investigated by e.p.r. spectroscopy of molybdenum(V). The signal obtained on extended reduction of the protein with sodium dithionite has been shown, by studies at 9 and 35 HGz in 1H2O and 2H2O and computer simulations, to have parameters corresponding to those of the Slow signal from the inactive desulpho form of various molybdenum-containing hydroxylases. Another signal obtained on brief reduction of the protein with small amounts of dithionite was shown by e.p.r. difference techniques to be a Rapid type 2 signal, like that from the active form of such enzymes. In confirmation that the protein is a molybdenum-containing hydroxylase, activity measurements revealed that it had aldehyde:2,6-dichlorophenol-indophenol oxidoreductase activity. No such activity towards xanthine or purine was observed. Salicylaldehyde was a particularly good substrate, and treatment of the protein with it also gave rise to the Rapid signal. Molybdenum cofactor liberated from the protein was active in the nit-1 Neurospora crassa nitrate reductase assay. It is concluded that the protein is a form of an aldehyde oxidase or dehydrogenase. From the intensity of the e.p.r. signals and from enzyme activity measurements, 10-30% of the protein in the sample examined appeared to be in the functional form. The evolutionary significance of the protein, which may represent a primitive form of the enzyme rather than a degradation product, is discussed briefly.