Discharge information for all Hong Kong government hospitals, which is routinely collected through the Clinical Management System (CMS), was used to assess the relative importance of all causes of diarrhoeal illness and to address the issue of under-diagnosis of rotavirus by linking discharge diagnostic codes with actual laboratory results for one hospital. Of all children less than 5 years of age hospitalized in Hong Kong in the 2-year period July 1997 to June 1999, 12257 (11%) were discharged with a primary diarrhoea diagnosis (74% coded as non-specified, 10·4% as rotavirus, 11% as Salmonella and 5% as other viral or bacterial). Linked laboratory and discharge data for one hospital demonstrated that 15% (n=1522) of all admissions had a primary diarrhoea diagnosis and that 40% of these had a specimen sent for rotavirus testing, of which 37% were positive. However, 46% (67/145) of children with a diagnosis of rotavirus infection had no virology result, and 69% (172/248) of positive rotavirus results were in children with no diagnosis indicating rotavirus infection. Modification of the CMS to routinely combine existing computerized laboratory data with the CMS discharge diagnoses and to develop mechanisms to enhance reliability of discharge diagnosis coding could produce a powerful resource for disease surveillance, auditing and for monitoring the impact of future vaccination and other prevention programmes.