Two previous studies on the mechanism of tyrosinase have given quite conflicting results. In a QM-only study using a rather small model, a mechanism was suggested in which the tyrosine proton is removed before catalysis. This was followed by catalytic cycles where a superoxo ligand attacks the phenolate ring. In another, more recent study, at the QM/MM level including the entire protein in the model, a quite different mechanism was instead advocated where a bridging O(2)H ligand was homolytically cleaved. That mechanism was rejected in the earlier QM-only study as having a prohibitively large barrier for O-O bond cleavage. In the present study, this discrepancy between the previous studies is investigated by new QM-only and QM/MM calculations.