Effective, computer-based representation of clinical observations requires balancing the advantages of structured, coded descriptions against those of free text narrative. An essential data set of relevant signs and symptoms was defined by a multidisciplinary group based on management goals published in a national guideline to meet the needs of clinicians in the Spina Bifida Clinic at Yale-New Haven Hospital. These core data elements are stored in a structured format. Additional material is stored as free text. A relational schema was devised that permits storage of both coded findings and narrative. Symptoms and signs are represented as subtypes of a supertype patient finding entity; they inherit common attributes and specialize others. The IVORY vocabulary was supplemented and modified to provide terms that describe relevant clinical observations. For this application, fields were added that enable predictive data entry of findings based on patient age and gender.