To be able to obtain useful information from medical data easily and quickly in daily use for the hospital staff, it is necessary to construct a proper database system using methodologies suitable for the nature of medical data. From this viewpoint we have developed the clinical test relational database system at Hiroshima University Hospital, which provides the functions of easy on-line retrieval and statistical analysis. Fuzzy query processing based on the fuzzy set theory is adopted in the system. This enables us to use natural linguistic representation and makes it easy to introduce medical knowledge representation. Fuzzy set approaches turn out to be superior in clinical evaluation of laboratory data to the ordinary clear-cut definition of normalcy.