The objective of this randomized study was to compare the occurrence of nosocomial pneumonia in nasotracheally intubated patients who were randomly allocated either to a systematic search of sinusitis by CT scan (study group) or not (control group). A total of 399 patients were included: 272 male and 127 female; mean age, 61 +/- 17 yr; SAPS: 12.6 +/- 4.9. The study group consisted of 199 patients and the control group consisted of 200. In the study group, sinus CT scans were performed in case of fever at Days 4 and 8 and then every 7 d. Nosocomial sinusitis was defined as follows: fever of >/= 38 degrees C, radiographic (sinusal air-fluid level or opacification on CT scan) signs, and presence of purulent aspirate from the involved sinus puncture with >/= 10(3) cfu/ml. Patients with sinusitis received sinus lavage and intravenously administered antibiotics. In the study group, 80 patients experienced nosocomial sinusitis. In the control group, no patient was treated for a sinusitis. Ventilator-associated bronchopneumonia (VAP) was observed in 88 patients: 37 in the study group (1 mo Kaplan-Meier estimate, 34%) versus 51 in the control group (1 mo Kaplan-Meier estimate, 47%); (p = 0.02, log-rank test; relative risk [RR] = 0.61; 95% confidence interval [CI], 0.40 to 0.93). Two months overall mortality was estimated at 36% in the study group versus 46% in the control group (p = 0.03, log-rank test; RR = 0.71; 95% CI, 0.52 to 0.97). We conclude that the occurrence of VAP in patients undergoing prolonged mechanical ventilation via a nasotracheal intubation can be prevented by the systematic search and treatment of nosocomial sinusitis. The effect on mortality should be confirmed.