Summaryobjective To identify useful predictors of morbidity of young children in central Africa. method Population-based follow-up study in Northern Kivu, Congo, of 842 children under two years of age who completed weekly follow-up interviews and health examinations during three months. Main outcome measures were crude and adjusted effects of summary measures of nutritional status on one-month cumulative incidence of malaria, respiratory illness, and diarrhoea.results Anthropometric indicators appeared to perform badly in predicting morbidity. In contrast, nonanthropometric variables such as growth as judged by the caretaker, child's diet at the time of examination, and occurrence of disease in the month preceding the interval of observation were useful.conclusions In the context of the 'Sick Child Initiative', simple tests and diagnostic tools to improve quality of both prevention and cure in first-level facilities need to be identified. Focusing on non-anthropometric indicators should be encouraged to offer a comprehensive appraisal of health status to all children.