The recovery pattern and outcome were analysed in 261 consecutive children (age 6-36 months) with persistent diarrhoea who underwent inpatient nutritional rehabilitation with a rice-lentil (Khitchri) and yoghurt-based diet. Overall, 217 (83%) recovered successfully, as judged by a reduction in stool output and weight gain for a consecutive 3 d. Failures were more commonly febrile at admission [odds ratio (OR) 2.3, 95% confidence interval (CI) 1.1-4.8] and a greater number had culture-proven sepsis (Fisher's exact test, p < 0.001). Logistic regression analysis identified significantly increased risk of treatment failure with several admission characteristics, including stool frequency > 5 d(-1) (OR 2.9, 95% CI 1.6-5.2), vomiting (OR 2.5, 95% CI 1.1-5.7) and sepsis (OR 2.8, 95% CI 1.1-7.5). Survival analysis revealed significantly longer time-to-recovery among children with stool frequency > 5 d(-1) at admission (p < 0.001), suspected sepsis necessitating intravenous antibiotics (p < 0.001) or oral candidiasis (p < 0.05). These findings suggest that severity of diarrhoea and coexisting systemic infections are key determinants of the response to nutritional therapy in children with persistent diarrhoea.