Ascorbate oxidase, dissolved in Hepes or sodium phosphate buffers, was analyzed by EPR and activity measurements before and after storage at -30 degrees C and 77 K. The specific activity was somewhat higher in the phosphate buffer, about 3500-3700 Dawson units compared to about 3100 units of the enzyme dissolved in Hepes buffer. After storage at -30 degrees C the activity fell to 1400-2000 units in the phosphate buffer but only to 2600-2800 units in the Hepes buffer. Large changes occurred in the EPR spectrum of enzyme dissolved in the phosphate buffer after storing at -30 degrees C suggesting an alteration of the type 2 copper site. These changes were, however, reverted when the samples were thawed and rapidly frozen at 77 K. Copper analysis showed that about 50% of the total copper was EPR detected. The type 2 Cu2+ EPR intensity was in most samples close to 25% of the total EPR intensity. This low contribution of type 2 Cu2+ could not be changed if the enzyme was completely reduced and reoxidized, treated with Fe(CN)6(3), large excess of NaF, addition of 50% (v/v) ethylene glycol or dialyzed against 0.1 M Mes buffer (pH 5.5). Since the crystal structure shows that there are one each of types 1 and 2 copper in the monomers there must be another species with an EPR signal rather different from these two copper species. This signal is proposed to originate from some trinuclear centers. The EPR simulations show that it is possible to house a broad unresolved signal under the resolved type 1 and 2 signals so that the total integral becomes 50% of the total copper in the molecule.