When aqueous solutions of dextran and of poly-(ethylene glycol) are mixed above certain concentrations, immiscible, liquid two-phase systems are obtained which are useful for separating cells by partition. Some salts partition unequally between the phases, giving rise to an electrostatic potential difference between them. Partition of cells has therefore been thought to depend predominantly on membrane charge. We now report two instances in which membrane charge either does not determine or is not the main determinant of cell partition. (A) Cell partition coefficients in phase systems approaching the critical point (the component concentrations below which a homogeneous solution occurs) increase, even in phase systems in which the phase potential difference is practically zero. Furthermore, in such systems, the partition coefficient of (human) erythrocytes is not reduced by complete removal of sialic acid. (B) Rat and mouse erythrocytes have sizable partition coefficients in a phase system away from the critical point with no potential difference between the phases. Cell surface interaction with the polymers is probably responsible for cell partition in these cases. Partition studies on erythrocytes from nine mammalian species in phases near the critical point with and without electrostatic potential differences reveal major species-specific differences in the membrane charge/noncharge components. A correlation has been found, in phases near the critical point that have essentially no electrostatic potential difference, between partition coefficient and the ratio of poly/monounsaturated fatty acids in the membranes of red cells from different species. Our present results thus provide parameters for the separation of cells by partition in addition to or instead of membrane charge depending on the polymer and salt composition and concentration selected.