A community-based longitudinal study was carried out at Matlab, Bangladesh, to investigate the magnitude of the problem of persistent diarrhea; 705 children aged < 5 y were followed, yielding 7300 child-months of observation in 1 y. Morbidity data were collected every fourth day by home visit, anthropometric status was determined monthly, and cell-mediated immune status was assessed every 3 mo. The incidence of persistent diarrhea was 34 episodes per 100 child-years; rates were highest in infancy and declined through the remainder of childhood. In a logistic-regression model, weight-for-height status and immune status were significant predictors of persistent diarrhea. Compared with those at zero Z score, children with weight-for-height at -2 would have a 3.5 times increased risk of persistent diarrhea. Compared with immunocompetent children, immunodeficient children had about twice the risk of developing persistent diarrhea. Thus, nutritional status and cell-mediated immune status were important independent risk factors for persistent diarrhea.