Xanthine oxidase is stable and active in aqueous dimethyl sulphoxide solutions of up to at least 57% (w/w). Simple techniques are described for mixing the enzyme in this solvent at--82 degrees C, with its substrate, xanthine. When working at high pH values under such conditions, no reaction occurred, as judged by the absence of e.p.r. signals. On warming to--60 degrees C, for 10 min, however, the Very Rapid molybdenum(V) e.p.r. signal was obtained. This signal did not change on decreasing the pH, while maintaining the sample in liquid nitrate reductase, caused its molybdenum(V) e.p.r. signal to change from the high-pH to the low-pH form. These findings are not compatible with the conclusions of Edmondson, Ballou, Van Heuvelen, Palmer & Massey [J. Biol. Chem. (1973) 248, 6135-6144], that the Very Rapid signal is in prototropic equilibrium with the Rapid signal, and should be important in understanding the mechanism of action of the enzyme. They emphasize the unique nature of the intermediate represented by the Very Rapid e.p.r. signal. The possible value of the pK for loss of an exchangeable proton from the Rapid signal is discussed.