To evaluate the attributable morbidity and mortality of ventilator-associated pneumonia (VAP) in intensive care unit (ICU) patients, we conducted a prospective, matched cohort study. Patients expected to be ventilated for > 48 h were prospectively followed for the development of VAP. To determine the excess ICU stay and mortality attributable to VAP, we matched patients with VAP to patients who did not develop clinically suspected pneumonia. We also conducted sensitivity analyses to examine the effect of different populations, onset of pneumonia, diagnostic criteria, causative organisms, and adequacy of empiric treatment on the outcome of VAP. One hundred and seventy-seven patients developed VAP. As compared with matched patients who did not develop VAP, patients with VAP stayed in the ICU for 4.3 d (95% confidence interval [CI]: 1.5 to 7. 0 d) longer and had a trend toward an increase in risk of death (absolute risk increase: 5.8%; 95% CI: -2.4 to 14.0 d; relative risk (RR) increase: 32.3%; 95% CI: -20.6 to 85.1%). The attributable ICU length of stay was longer for medical than for surgical patients (6. 5 versus 0.7 d, p < 0.004), and for patients infected with "high risk" organisms as compared with "low risk" organisms (9.1 d versus 2.9 d). The attributable mortality was higher for medical patients than for surgical patients (RR increase of 65% versus -27.3%, p = 0. 04). Results were similar for three different VAP diagnostic criteria. We conclude that VAP prolongs ICU length of stay and may increase the risk of death in critically ill patients. The attributable risk of VAP appears to vary with patient population and infecting organism.