The nature of the primary lesions suffered by cells during freezing and thawing is unclear, although the plasma membrane is often considered the primary site for freezing injury. This study was designed to investigate the nature of damage immediately after thawing, by monitoring several functional tests of the cell and the plasma membrane. Hamster fibroblasts, human lymphocytes, and human granulocytes were subjected to a graded freeze-thaw stress in the absence of cryoprotective compound by cooling at -1 degree C/min to a temperature between -10 and -40 degrees C, and then were either warmed directly in water at 37 degrees C or cooled rapidly to -196 degrees C before rapid warming. Mitochondrial function in the cells was then assessed using 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl-tetrazolium bromide (MTT), fluorescein diacetate (FDA), colony growth, and osmometric response in a hypertonic solution. Cells behaved as osmometers after cooling at -1 degree C/min to low temperatures at which there were no responses measured by other assays, indicating that the plasma membrane is not a primary site for injury sustained during slow cooling. These results also indicate that the FDA test does not measure membrane integrity, but reflects the permeability of the channels through which fluorescein leaves the cells. Fewer cells could respond osmotically after cooling under conditions where intracellular freezing was likely, implying that the plasma membrane is directly damaged by the conditions leading to intracellular freezing. A general model of freezing injury to nucleated mammalian cells is proposed in which disruption of the lysosomes constitutes the primary lesion in cells cooled under conditions where the cells are dehydrated at low temperatures.