To describe the epidemiologic and microbial aspects of ventilator-associated pneumonia (VAP) in patients with acute respiratory distress syndrome (ARDS), we prospectively evaluated 243 consecutive patients who required mechanical ventilation (MV) for > or = 48 h, 56 of whom developed ARDS as defined by a Murray lung injury score > 2.5. We did this with bronchoscopic techniques when VAP was clinically suspected, before any modification of existing antimicrobial therapy. For all patients, the diagnosis of pneumonia was established on the basis of culture results of protected-specimen brush (PSB) (> or = 10(3) cfu/ml) and bronchoalvelolar lavage fluid (BALF) (> or = 10(4) cfu/ml) specimens, and direct examination of cells recovered by bronchoalveolar lavage (BAL) (< or = 5% of infected cells). Thirty-one (55%) of the 56 patients with ARDS developed VAP for a total of 41 episodes, as compared with only 53 (28%) of the 187 patients without ARDS for a total of 65 episodes (p = 0.0005). Only 10% of first episodes of VAP in patients with ARDS occurred before Day 7 of MV, as compared with 40% of the episodes in patients without ARDS (p = 0.005). All but two patients with ARDS who developed VAP had received antimicrobial treatment (mostly with broad-spectrum antibiotics) before the onset of infection, as compared with only 35 patients without ARDS (p = 0.004). The organisms most frequently isolated from patients with ARDS and VAP were methicillin-resistant Staphylococcus aureus (23%), nonfermenting gram-negative bacilli (21%), and Enterobacteriaceae (21%). These findings confirm that microbiologically provable VAP occurs far more often in patients with ARDS than in other ventilated patients. Because these patients are often treated with antibiotics early in the course of the syndrome, the onset of VAP is frequently delayed after the first week of MV, and is then caused mainly by methicillin-resistant S. aureus and other multiresistant microorganisms.