Multiple freeze-thaw (FT) cycles, with complete melting between cycles, resulted in an exponential decline in liver alcohol dehydrogenase (LADH) enzyme activity. The reduction in activity of LADH as a result of FT damage was proportional to the decrease in the intensity of the tryptophan fluorescence of the enzyme. Treatment with urea resulted in a similar relationship between tryptophan fluorescence intensity and inactivation. Evidence from fluorescence and activity studies from the same sample, as well as gel electrophoresis, indicates that damage to LADH from a FT cycle, resulting in inactivation, is likely an unfolding of the enzyme rather than separation of subunits or aggregation of enzymes at the enzyme concentrations and cooling rates used. A nonexponential decline in enzyme activity, as a function of the number of FT cycles, can be achieved if complete melting between cycles is not allowed or if the samples are stored at +4 degrees C for 24 hr following the last FT cycle, prior to assay. In the latter case, a partial recovery in enzyme activity is seen. "Seeding," while lowering the enzyme activity, is desirable to achieve consistent results without the artifacts that are introduced if not used. Amino acids were tested for their effectiveness as cryoprotectants. From the results of this study, the mean fractional area loss of amino acid residues upon incorporation in globular proteins (f) is inversely proportional to the FT protection by these free amino acids. Thus, amino acid residues which tend to be found at the surface of proteins (e.g., glutamate) improve the FT survival of LADH, when added as the free amino acid, while those amino acids which are found in the interior of proteins (e.g., valine, leucine) sensitize LADH to FT damage. The pattern of protection ("fingerprint") of LADH by various amino acids is different from that of living cells. Furthermore, unlike the case with cells, glutamine and DMSO do not act independently when protecting LADH.