The district general hospital (DGH) is a common feature of health service provision in many developing countries. We have used linked demographic and clinical surveillance in a rural community located close to a DGH on the Kenyan coast to define the use and public health significance of essential clinical services provided by it. Of a birth cohort of over 4000 children followed for approximately 6 years, about a third were admitted to hospital at least once. Significantly more children admitted with major infectious diseases such as malaria and acute respiratory tract infections were readmitted with the same condition during the surveillance period than would have been expected by chance. Among surviving admissions, mortality post-discharge was significantly higher than in the cohort which had not been admitted within 3, 6 and 12 months. Most of the patients who died after discharge had been admitted with a diagnosis of gastroenteritis. Most children admitted to the DGH survive hospitalization and the remaining period of childhood. Despite no clinical trial evidence to support the claim, it seems reasonable to assume that in the absence of intensive clinical management provided by a DGH, a significant proportion of these children would not have survived. However, the DGH is able to define a group of at-risk children who re-present with severe complications of infectious disease, and of these several may have underlying conditions not amenable to DGH intervention and continue to have a poor prognosis. Both groups of children represent statistically significant subsets of a rural paediatric community and the future organization and co-ordination of DGH and primary care services need to work in unison to strengthen the service needs of children at risk.