SUMMARY1. Human red cells treated with lactose solution and loaded with NaCl and BCI subsequently exchange cation with a nutrient BCl medium. B is the same in cells and medium, and is either K, Rb, Cs or Li. In these circumstances Na always moves outwards with the concentration gradient, but the efflux is largely active.2. With suspensions in media containing Ca2 , the total Na efflux depends on the amount of Na in the cells and on the nature of cation B. Thus for any given value of mean cell Na (Nam) in excess of 30 m-equiv/l. cells, the effect of B on the amount of Na efflux is K > Cs > Rb > Li, while with Nam between 0 7 and 5 m-equiv/l. cells, the sequence is Cs > Li > K, Rb. With Nam between 5 and 30 m-equiv/l. cells, intermediate sequences may be demonstrated for the effects of B on Na efflux. This applies both to the efflux itself and to the flux: concentration ratio, FCR.3. FCR for passive Na efflux in these circumstances is determined by adding strophanthin G to the medium. It varies inversely with the duration of exposure to Ca2+ in the exchange and nutrient media, but not with the nature of cation B. FCR for passive efflux is probably little affected by the value of Nam.4. By deducting passive from total Na efflux, active Na efflux is obtained, and variations in the latter with cell Na content, and with B, result in FCR curves similar to those obtained with total Na efflux.5. Total and passive Na efflux have also been measured in Ca-free media. Here the passive efflux is considerable, and with Na + K cells in KCI media FCR increases with Nam, but in other systems this change is not signi-* In receipt of a personal grant from the Wellcome Trust, and a grant for equipment from the Public Health Service, U.S.A. (HE 08817-01) (National Heart Institute). ficant. However, the rate of passive efflux into LiCl media is less than that for KCO or CsCl media. Owing to the magnitude of passive flux in Ca-free systems, the total Na efflux is also increased, but FCR for active Na efflux is quantitatively and qualitatively similar to that occurring in systems containing Ca2+.6. The effects of B and Nanm on Na efflux give a series of sequences for B which recall some of those obtainable when chemically modified glass membranes separate solutions of salts, and which are attributable to the charge on the membrane and the hydration of the cations involved.