A new computer-assisted infection monitoring (CAI) software program has been developed for use in an intensive-care unit (ICU). By means of an interactive dialogue with physicians at the bedside, infection diagnoses and therapeutic decisions were recorded prospectively during a 3-month test period. By linking epidemiological data with information about therapeutic decisions, CAI could assess the quality of the therapeutic decisions. Antibiotics chosen empirically before the availability of any culture results, matched the antibiotic susceptibility patterns of the subsequently identified pathogens in 74% of the cases. Therapy chosen in collaboration with the computer after the pathogen was known, but before sensitivity results were available, corresponded with the eventual antibiograms of the microorganisms in 90% of the cases. Data analysis by CAI allowed us to assess critically the diagnostic and therapeutic habits in our ICU. Using the query-by-example method, CAI automatically calculated device-associated infection rates.