One of the remaining mysteries regarding the respiratory enzyme cytochrome c oxidase is how proton pumping can occur in all reduction steps in spite of the low reduction potentials observed in equilibrium titration experiments for two of the active site cofactors, CuB(II) and Fea3(III). It has been speculated that, at least the copper cofactor can acquire two different states, one metastable activated state occurring during enzyme turnover, and one relaxed state with lower energy, reached only when the supply of electrons stops. The activated state should have a transiently increased CuB(II) reduction potential, allowing proton pumping. The relaxed state should have a lower reduction potential, as measured in the titration experiments. However, the structures of these two states are not known. Quantum mechanical calculations show that the proton coupled reduction potential for CuB is inherently high in the active site as it appears after reaction with oxygen, which explains the observed proton pumping. It is suggested here that, when the flow of electrons ceases, a relaxed resting state is formed by the uptake of one extra proton, on top of the charge compensating protons delivered in each reduction step. The extra proton in the active site decreases the proton coupled reduction potential for CuB by almost half a volt, leading to agreement with titration experiments. Furthermore, the structure for the resting state with an extra proton is found to have a hydroxo-bridge between CuB(II) and Fea3(III), yielding a magnetic coupling that can explain the experimentally observed EPR silence.