Upon reaction of cytochrome oxidase with hydrogen peroxide, the spectral changes are complete, with slightly less than 1 equiv of hydrogen peroxide per cytochrome oxidase. At pH 8 the product is a mixture of the P and F forms, while at pH 6 the product is exclusively the F form. These data are inconsistent with current interpretations of the structure of compounds P and F. Two stable radical species are detected by EPR; the relative amounts of these species are pH dependent. The MCD spectra of pure P and F are reported. It is suggested that compound F is a hydrogen peroxide adduct of cytochrome oxidase with cytochrome a3 in the low-spin state and that compound P is an oxyferryl state of cytochrome alpha 3 in support of the recent Raman data of Proshlyakov et al. [(1994) J. Biol. Chem. 269, 29385-29388]. We also suggest that copper B is in the trivalent state in compound P.