OBJECTIVE: to investigate the epidemiology of infection by Pseudomonas aeruginosa in patients with ventilator-associated pneumonia (VAP). DESIGN: prospective clinical study. SETTING: a medical-surgical ICU in a university hospital. PATIENTS: we followed-up 568 mechanically ventilated patients and 83 episodes of VAP with etiologic diagnosis in 72 patients were retained for analysis. RESULTS: Ps. aeruginosa was isolated in 22 (26.5%) episodes in 18 patients. Of these episodes 7 were directly responsible for death. Using logistic regression analysis, the risk of VAP due to Ps. aeruginosa was increased in patients with chronic obstructive pulmonary disease (relative risk (RR) = 29.9, 95% confidence interval (CI) = 4.86-184.53), a mechanical ventilation period longer than 8 days (RR = 8.1, 95% CI = 1.01-65.40) and prior use of antibiotics (RR = 5.5, 95% CI = 0.88-35.01). CONCLUSIONS: patients with VAP and these factors have a greater risk of infection by Ps. aeruginosa and empirical therapy for these episodes should include anti-pseudomonal activity until etiologic diagnosis is established.