One widely accepted explanation of injury from slow freezing is that damage results when the concentration of electrolyte reaches a critical level in partly frozen solutions during freezing. We have conducted experiments on human red cells to further test this hypothesis. Cells were suspended in phosphate-buffered saline containing 0-3 M glycerol, held for 30 min at 20 degrees C to permit solute permeation, and frozen at 0.5 or 1.7 degrees C/min to various temperatures between -2 and -100 degrees C. Upon reaching the desired minimum temperature, the samples were warmed at rates ranging from 1 to 550 degrees C/min and the percent hemolysis was determined. The results for a cooling rate of 1.7 degrees C/min indicate the following: (a) Between 0.5 and 1.85 M glycerol, the temperature yielding 50% hemolysis (LT50) drops slowly from -18 to -35 degrees C. (b) The LT50's over this range of concentrations are relatively independent of warming rate. (c) With glycerol concentrations of 1.95 and 2.0 M, the LT50 drops abruptly to -60 degrees C and to below -100 degrees C, respectively, and becomes dependent on warming rate. The LT50 is lower with slow warming at 1 degree C/min than with rapid. With still higher concentrations (2.5 and 3.0 M), there is no LT50, i.e., more than 50% of the cells survive freezing to-100 degrees C. Results for cooling at 0.5 degrees C/min in 2 M glycerol were similar except that the LT50s were some 10-20 degrees C higher. A companion paper (Rall et al., Biophys. J. 23:101-120, 1978) examines the relation between survival and the concentrations of salts produced during freezing.