To determine the effect of nutritional state on persistent diarrhoea a case-control study was carried out on 756 children followed up prospectively for 18 months. Children who developed persistent diarrhoea were compared with population controls and controls with acute diarrhoea. The mean weight for age in the children with persistent diarrhoea (69.9%) was significantly lower than that in the population controls (77.0%) and the diarrhoeal controls (76.2%). Weight for age of less than or equal to 70% was associated with persistent diarrhoea in both case-control analyses (population controls, matched odds ratio 3.25; diarrhoeal controls, matched odds ratio 2.46). The corrected odds (multiple logistic regression) in the two analyses were 3.2 (95% confidence interval 1.3 to 8.1) and 3.4 (1.2 to 9.1). Weight for age of less than or equal to 70% increases the risk of persistent diarrhoea. In an underweight child there is a higher risk of diarrhoea becoming persistent. Prevention of malnutrition and intensive management of acute diarrhoea in malnourished children should help reduce the risk of the diarrhoea persisting.