A prospective observational cohort study was undertaken with two endpoints: (1) to compare the time cut-off of 48h and the carrier state criterion for classifying lower airway infections in adult and paediatric long-term ventilated patients, and (2) to evaluate the potential of optimized time cut-offs for characterizing imported and ICU-acquired lower airway infections. All patients admitted to the general and paediatric intensive care units and expected to require mechanical ventilation for a period > or = 3 days were enrolled. Surveillance cultures of throat and rectum were obtained on admission and thereafter twice weekly to distinguish micro-organisms that were imported into the unit from those acquired during the stay on the unit. A total of 130 adults and 400 children were studied. In the adult population, 70% of lower airway infections were classified as ICU-acquired by the 48 h cut-off and 48% by the criterion of carriage; on the paediatric ICU the percentages were 65% and 20%, respectively. To separate imported from ICU-acquired infections, eight days was optimal in the adult population and 10 days in the paediatric population. Sensitivity, specificity, positive predictive value and negative predictive value for a time cut-off of eight days for adults were 86, 77, 80, 83%, respectively, and using 10 days for children were 87, 62, 90, 56%, respectively. The use of the 48 h cut-off rule classifies patients as having nosocomial pneumonia, when in fact the infections are commonly caused by microorganisms carried in by the patients. In contrast, using the carriage method, the proportion of lung infections due to nosocomial bacteria was relatively small and was a late phenomenon. Although in prolonging the time cut-off the difference between the two types of classification was shorter, time cut-offs were still found to be unreliable for distinguishing imported from unit-acquired lower airway infections.