Ventilator-associated pneumonia (VAP) is a common problem in an intensive care unit (ICU), although the incidence is not well established. This study aims to compare the VAP incidence as determined by the treating surgical intensivist with that detected by the hospital Infection Control Service (ICS). Trauma and surgical patients admitted to the surgical critical care service were prospectively evaluated for VAP during a 5-month time period. Collected data included the surgical intensivist's clinical VAP (SIS-VAP) assessment using Centers for Disease Control and Prevention (CDC) VAP criteria. As part of the hospital's VAP surveillance program, these patients' medical records were also reviewed by the ICS for VAP (ICS-VAP) using the same CDC VAP criteria. All patients suspected of having VAP underwent bronchioalveolar lavage (BAL). The SIS-VAP and ICS-VAP were then compared with BAL-VAP. Three hundred twenty-nine patients were admitted to the ICU during the study period. One hundred thirty-three were intubated longer than 48 hours and comprised our study population. Sixty-two patients underwent BAL evaluation for the presence of VAP on 89 occasions. SIS-VAP was diagnosed in 38 (28.5%) patients. ICS-VAP was identified in 11 (8.3%) patients (P < 0.001). The incidence of VAP by BAL criteria was 23.3 per cent. When compared with BAL, SIS-VAP had 61.3 per cent sensitivity and ICS-VAP had 29 per cent sensitivity. VAP rates reported by hospital administrative sources are significantly less accurate than physician-reported rates and dramatically underestimate the incidence of VAP. Proclaiming VAP as a never event for critically ill for surgical and trauma patients appears to be a fallacy.