Previous studies of erythrocyte ion (potassium and sodium) transport during marasmus and kwashiorkor have indicated increased passive permeation to both ions in both syndromes, and increased Na,K pump activity in kwashiorkor and reduced activity in marasmus. Children with severe cerebral palsy (CP) frequently suffer secondary protein energy malnutrition (PEM). Unlike marasmus and kwashiorkor, this PEM is uncomplicated by micronutrient deficiency, parasitism and infections. Because of deformities classification of PEM cannot be performed in these children by stature-based anthropometry, therefore we used triceps skinfold thicknesses less than the fifth percentile and absence of weight gain in the previous year as criteria for malnutrition. K influx data from well- and malnourished CP children, and from well-nourished controls reveal that ouabain-sensitive K influx is highest in malnourished CP, followed by well-nourished CP (P = 0.02), and lowest in controls (P less than 0.001, vs. malnourished). Determinations of ouabain-sensitive Na efflux, though less precise and therefore more variable, were consistent with this finding of no decrease of Na,K pump activity occurring during the development of this malnutrition. There were no statistically significant differences in ouabain-insensitive fluxes of either Na or K. Ion transport in undernourished CP children thus resembles that found in kwashiorkor rather than in marasmus; but oedema is rarely seen in this form of secondary PEM.