A systematic study on the stability of frozen erythrocytes was performed. Washed and concentrated erythrocytes were mixed with an equal volume of cryoprotective solution containing 24% (w/w) hydroxyethyl starch (HES) and 60 mmol/liter NaCl according to an established protocol. Volumes of 250 microliters of this mixture were filled into polypropylene tubes and cooled to -196 degrees C with a rate of 293 degrees C/min by immersion in liquid nitrogen. The storage temperature was then varied from -10 to -75 degrees C and could be identified as the predominant factor influencing hemolysis kinetics. The effect of storage temperature on the frozen erythrocytes after thawing was evaluated by measuring the hemolysis in a dilute, isotonic NaCl solution (saline stability). A strong time dependence was found within the temperature range studied and could be described by an exponential kinetic law. A stability prediction was made for storage temperatures lower than those examined. Temperature ranges of qualitatively different hemolysis kinetics were identified and compared to devitrification behavior of intra-and extracellular solutions. The intracellular solution was simulated by a concentrated mixture of dried erythrocytes and water. The devitrification behavior was studied using DSC techniques. A rapidly frozen mixture was annealed at selected temperatures which fall into the range of storage temperatures for frozen erythrocytes. This paper tentatively interprets the devitrification data with respect to the means for cell damage during storage. The results are reviewed with respect to the design of a safe storage procedure.