BACKGROUND: Ventilator-associated pneumonia is a common cause of morbidity in critically ill patients. Interventions beneficial to the prevention of ventilator-associated pneumonia would therefore have a significant impact on the care of these patients. PURPOSE: To perform a literature review and synthesis of methods for prevention of ventilator-associated pneumonia. DATA SOURCES: MEDLINE (1966-2001), the Cochrane Library, and bibliographies of retrieved articles. STUDY SELECTION: Studies were required to be prospective and controlled in design and to evaluate clinically important or surrogate outcomes. Surrogate outcomes were required to have a direct link to clinically important outcomes supported by the literature. DATA EXTRACTION: Data on patients, definitions, study design, and outcomes were abstracted and graded by using preestablished criteria. DATA SYNTHESIS: The preventive practices with the strongest supportive evidence were semi-recumbent positioning, sucralfate instead of H2-antagonists for stress ulcer prophylaxis, and selective digestive tract decontamination. Aspiration of subglottic secretions and oscillating beds may be useful in select populations. There is no evidence to support specific methods of enteral feeding or increased frequency of ventilator circuitry changes. CONCLUSIONS: After evaluation of potential benefits and risks, the authors recommend considering several specific interventions to reduce the incidence of ventilator-associated pneumonia: semi-recumbent positioning in all eligible patients, sucralfate rather than H2-antagonists in patients at low to moderate risk for gastrointestinal tract bleeding, and aspiration of subglottic secretions and oscillating beds in select patient populations. Selective digestive tract decontamination is not recommended because routine use may increase antimicrobial resistance.