The purpose of this study was to investigate the applicability of the case-control method in the detection of relevant behavioral characteristics. This will aid in the design of interventions to reduce the occurrence of severe diarrhea among children under three years of age in the City of Kinshasa, Zaire Between March and November 1988, one hundred and seven patients with severe diarrhea were selected among children under the age three admitted at two hospitals in Kinshasa. An equal number of age-matched controls were selected in the neighborhood of the cases. Following case hospital discharge, structured interviews were conducted among case and control mothers, on selected hygiene and sanitation behaviors, knowledge of diarrhea causation, and health beliefs. These interviews were organized in the household settings. During the same day, a direct observation was performed on behavior and environmental sanitation Cases of severe diarrhea originated from twenty-one of the twenty-four administrative zones of Kinshasa. There were relatively more cases from the closer zones than from the more distant zones. Also, there were more cases from the poorer zones than from medium-income or affluent zones. Stool examination indicated presence of enteropathogenic E. Coli in thirteen percent of the case stools, shigella spp in six percent, salmonella spp in four percent, and staphylococcus in three percent. Very few stool samples were examined for the presence of virus The unadjusted univariate analysis, using McNemar's chi-square test for matched pairs, identified thirteen risk factors relevant to knowledge, health beliefs, sanitation behavior, hygiene behavior, and child supervision. Multivariate analysis using logistic regression retained indicators of sanitation behavior, hygiene behavior and health beliefs as the only significant risk factors for severe diarrhea This study was able to pinpoint specific high risk behaviors for severe diarrhea in a community that is fairly homogeneous in cultural, environmental and socioeconomic characteristics. An approach of preventive measures is recommended, which takes into account the limited availability of resources in the studied population and the difficulties of the behavior-modification undertaking