The structure of histidine 94-->aspartate (H94D) carbonic anhydrase II (CAII) crystallized in an orthorhombic space group has been determined to 2.5 A resolution. This crystal form is not isomorphous with monoclinic wild-type enzyme crystals or with the monoclinic crystal form of H94D CAII reported earlier [Kiefer,L.L., Ippolito, J.A., Fierke, C.A. and Christianson, D.W. (1993) J. Am. Chem. Soc., 115, 12581-12582]. In monoclinic H94D CAII, a fully occupied zinc ion is tetrahedrally coordinated by D94, H96, H119 and a water molecule. In orthorhombic H94D CAII, a partially occupied zinc ion is coordinated by H96 and H119 and only weakly coordinated by a disordered D94 side chain. These differences are particularly surprising given that the two crystal forms co-precipitate in the same drop in the same experiment. Re-refinement of the orthorhombic crystal form of H94C CAII and comparison with its corresponding monoclinic crystal form yield similar results. It appears that partial-but not full-zinc dissociation accompanies the crystallization of CAII variants in the orthorhombic crystal form, and significant differences on the protein surface presumably affect the relative stability of each crystal lattice. These results underscore an unexpected ambiguity in this protein engineering experiment: which crystal structure of H94D CAII should be correlated with functional measurements made in solution?