The form of cytochrome d in Escherichia coli and Azotobacter vinelandii that shows an absorption maximum at 648 to 652 nm ('cytochrome d650') is generally regarded as the oxidized form of this terminal oxidase. Membranes from E. coli grown under oxygen-limited conditions, when treated with ferricyanide, do not reveal cytochrome d650, whereas a sharp symmetrical band at 652 nm results from the reaction of the reduced enzyme with O2 at either room temperature or after flash photolysis of the CO-liganded form at -130 degrees C. Electron paramagnetic resonance spectroscopy of cytochrome d650 trapped at -130 degrees C shows that its spectrum is indistinguishable from the CO-liganded form and does not reveal resonances of high spin ferric haem previously attributed to cytochrome d. An hypothesis is proposed in which cytochrome d650 is an early intermediate in the reaction of reduced cytochrome d650 and oxyhaemoglobin is presented and the hypothesis discussed in relation to earlier work, in which the indirect interconversions of reduced cytochrome d and d650 have been explained by proposing the existence of an 'invisible' form. It is suggested that this form could be the oxidized enzyme.