A study was made of sodium efflux from squid giant axon, and its sensitivity to external K and Na. When sodium efflux from untreated axons was strongly stimulated by Ko, Nao was inhibitory; when dependence on Ko was low, Nao had a stimulatory effect. Incipient CN poisoning or apyrase injection, which produces high intracellular levels of ADP1 and Pi, rendered sodium efflux less dependent on external K and more dependent on external Na. Injection of ADP, AMP, arginine, or creatine + creatine phosphokinase, all of which raise ADP levels without raising Pi levels, had the same effect as incipient CN poisoning. Pi injection had no effect on the K sensitivity of sodium efflux. Axons depleted of arginine and phosphoarginine by injection of arginase still lost their K sensitivity when the ATP:ADP ratio was lowered and regained it partially when the ratio was raised. Rough calculations show that sodium efflux is maximally Ko-dependent when the ATP:ADP ratio is about 10:1, becomes insensitive to Ko when the ratio is about 1:2, and is inhibited by Ko when the ratio is about 1:10. Deoxy-ATP mimicked ADP when injected into intact axons. Excess Mg, as well as Pi, inhibited both strophanthidin-sensitive and strophanthidin-insensitive sodium efflux. An outline is presented for a model which might explain the effects of ADP, Pi and deoxy-ATP.