The use of health diaries to monitor patients with chronic diseases has often been complicated by difficulties encountered in data quality assurance and interpretation. An expert system, Monitor, has been developed to predict the health status of cystic fibrosis patients based on daily home measurements of pulse, respiratory rate, weight, inspired vital capacity, and a check list of symptoms of acute illness. This system ensures data reliability beyond what can be achieved in most current automatic error detection procedures by validating inputs against patient-specific expectations. Its explicit representation of the time dimension and the hierarchical structure of its knowledge base facilitate the abstraction of trends and relationships among the time-dependent data. Dynamically imposed expectations also lend flexibility to the interpretation process by allowing the processing of partial (incomplete) data. Monitor correctly classified 86 percent (three-category classification) and 94 percent (two-category classification) of 111 cases. This demonstrates that expert systems can be a feasible approach in building more robust diary monitoring systems.