BACKGROUND: In order to achieve simultaneously a higher quality of hospital care and cost-containment, an objective assessment of alternative systems for surveillance of hospital infections is needed in our country. This assessment will allow to make a rational choice of the system to be used at each hospital. METHODS: 719 patients admitted to a Vascular Surgery ward were studied. Sensitivity, specificity, and predictive values for a selective system of surveillance of nosocomial infection (NI) were assessed. This selective system attempted to contain costs by means of limiting the revision of records to those patients who presented fever, prescription of antibiotics after admission or who required laboratory cultures. In this selective method charts revision was retrospectively performed by nurses. RESULTS: A low sensitivity (50.9%) and acceptable values for specificity and predictive values were found. CONCLUSIONS: Although this selective method requires only 20% of the reference method's time, its association with an important reduction in sensitivity precludes its systematic use and confines it eventually to low risk wards.