The role of dietary protein deficiency in kwashiorkor is uncertain, although it has been shown not to be involved in the famine oedema of adults. A study of six different diets given to 103 children with oedematous malnutrition showed that the rate of loss of oedema was strongly correlated with the dietary energy intake (r = 0.75) but not with the protein intake (r = 0.03). 66 patients given a very-low protein diet (2.5% protein energy) lost oedema as fast as those given five times as much protein. The energy intake above which oedema resolved and below which oedema accumulated was 245-270 KJ/kg/day. Because energy deficiency is not invariably associated with oedema it cannot be the only factor involved, and the necessary dietary component(s) must therefore have been present in surfeit in all the therapeutic diets. This could be potassium together with factors necessary for its retention. The accessory ingredients must be low in foods associated with human and experimental nutritional oedema. It is suggested that protein deficiency is not the cause of the oedema of kwashiorkor and that there is no need to postulate a different pathogenesis for this oedema from starvation oedema of adults.