The implications of the substitution of nickel for cobalt in Co,O, to give Co,NiO, are discussed. From the effects of substitution of other ions into Co,NiO, its curious properties may partly be understood. The cations present are unable CO satisfy the requirements ol both crystallographic siting and oxidation state to fit the classical spine1 distribution. The result is a weakening of the crystal lattice and hence rather loosely bonded oxygen, eg possibly as O-instead of 02-. This factor causes the low thermal stability and the exceptional activity towards oxygen reduction, and is maximised by incorporating the highest possible proportion of nickel and then heat-treating the catalyst near its point of breakdown. The importance of preparing the oxide in complete absence of OHor Hz0 is deduced. A site and charge distribution for Co? NiO, based on this work and other published data is advanced. The importancc of the fcrrimapnetic order remains to be clarilied. The electrocatalyst is gradually and il-reversibly deactivated on exposure to alkalme electrolyte. A mechantsm for this unfortunate property is suggested.