In 58 previously well-nourished patients who were fed cow's milk and who later developed persistent diarrhoea (greater-than-or-equal-to 15 days), data obtained during the first 8 days of acute diarrhoea were compared with those of patients whose episodes lasted less-than-or-equal-to 8 days. Children with persistent diarrhoea weighted less at birth, passed greater-than-or-equal-to 6 stools/day during the first 48 hours, received early treatment with antibiotics. Their milk feedings were stopped during the first 48 hours, they had a past history of digestive diseases and hospitalizations, they were brought in for consultation after 5 days of symptoms and their nutritional status deteriorated more (p < 0.003, p < 0.03, p < 0.0001, p < 0.0001, p < 0.0001, p < 0.002, p < 0.0001, and p < 0.03, respectively). Their mothers were significantly younger (p < 0.0013), had better schooling (p < 0.037), and fewer children (p < 0.044), and were separated from the index child during the day more often (p < 0.056). After persistent diarrhoea was diagnosed, enteropathogens in stools or lactose intolerance or both were demonstrated in 75.9%. Treatment induced remission in all cases. Using logistic regression a predictive model was established which enables us to identify, among patients with acute diarrhoea, those at risk of prolonging their illness.