The hematocrit method as a technique for determining red cell volume under anisotonic conditions has been reexamined and has been shown, with appropriate corrections for trapped plasma, to provide a true measure of cell volume. Cell volume changes in response to equilibration in anisotonic media were found to be much less than those predicted for an ideal osmometer; this anomalous behavior cannot be explained by solute leakage or by the changing osmotic coefficient of hemoglobin, but is quantitatively accounted for by the hypothesis that 20 per cent of intracellular water is bound to hemoglobin and is unavailable for participation in osmotic shifts.