Although copper is quantitatively removed from fungal laccase (Polyporus versicolor) by extended dialysis against high concentrations of cyanide, we have been unable to reconstitute the protein by addition of Cu(I) ions. However, two new methods for reversibly removing the type 2 Cu centre have been developed. The visible absorption at 610 nm, which is attributable to type 1 Cu, is unaffected by the procedure, but the absorbance of the type 3 Cu at 330 nm is decreased by 60 +/- 10%. The decrease is due, at least in part, to partial reduction of the binuclear type 3 centre, although there may be some change in the molar absorptivity of the oxidized chromophore as well. The change in the c.d. spectrum that occurs at approx. 350 nm may be explained in the same way, but it may also reflect the loss of a signal due to the type 2 Cu. Upon removal of the type 2 Cu an absorbance increase appears at approx. 435 nm, and it is assigned to the semi-reduced form of the type 3 pair. In the e.p.r. spectrum of the type 2-depleted enzyme the type 1 Cu signal exhibits well-resolved ligand hyperfine splitting, which can be simulated on the basis of contributions from two N and two H nuclei (AH congruent to AN congruent to 25 MHz). The H atoms are assumed to be attached to the beta-carbon of the covalently bonded cysteine ligand. A signal from a semi-reduced form(s) of the type 3 site can also be resolved in the spectrum of the type 2-depleted enzyme, and on the basis of the second integral of the e.p.r. spectrum 40% of the type 3 pairs are believed to be in a partially reduced state. The semi-reduced type 3 site is remarkably stable and is not readily oxidized by H2O2 or IrCl6(2-) or reduced by Fe(CN)6(4-). Intramolecular electron transfer is apparently quite slow in at least some forms of the type 2-depleted enzyme, and this may explain why the activity is at best 5% of that of the native enzyme. Full activity returns when type 2 copper is restored.