The phenomenon of intracellular freezing in cells was investigated by designing experiments with cultured mouse fibroblasts on a cryomicroscope to critically assess the current hypotheses describing the genesis of intracellular ice: (a) intracellular freezing is a result of critical undercooling; (b) the cytoplasm is nucleated through aqueous pores in the plasma membrane; and (c) intracellular freezing is a result of membrane damage caused by electrical transients at the ice interface. The experimental data did not support any of these theories, but was consistent with the hypothesis that the plasma membrane is damaged at a critical gradient in osmotic pressure across the membrane, and intracellular freezing occurs as a result of this damage. An implication of this hypothesis is that mathematical models can be used to design protocols to avoid damaging gradients in osmotic pressure, allowing new approaches to the preservation of cells, tissues, and organs by rapid cooling.