Oxidised cytochrome c oxidase is known to react with two molecules of hydrogen peroxide to form consecutively 607 nm 'Peroxy' and 580-nm 'Ferryl' species. These are widely used as model compounds for the equivalent P and F intermediates of the catalytic cycle. However, kinetic analysis of the reaction with H(2)O(2) in the pH range 6.0-9.0 reveals a more complex situation. In particular, as the pH is lowered, a 580-nm compound can be formed by reaction with a single H(2)O(2). This species, termed F(&z.rad;), is spectrally similar, but not identical, to F. The reactions are equivalent to those previously reported for the bo type quinol oxidase from Escherichia coli (T. Brittain, R.H. Little, C. Greenwood, N.J. Watmough, FEBS Lett. 399 (1996) 21-25) where it was proposed that F(&z.rad;) is produced directly from P. However, in the bovine oxidase F(&z.rad;) does not appear in samples of the 607-nm form, P(M), produced by CO/O(2) treatment, even at low pH, although this form is shown to be identical to the H(2)O(2)-derived P state, P(H), on the basis of spectral characteristics and kinetics of reaction with H(2)O(2). Furthermore, lowering the pH of a sample of P(M) or P(H) generated at high pH results in F(&z.rad;) formation only on a minutes time scale. It is concluded that P and F(&z.rad;) are not in a rapid, pH-dependent equilibrium, but instead are formed by distinct pathways and cannot interconvert in a simple manner, and that the crucial difference between them lies in their patterns of protonation.