The role of anaerobic bacteria in ventilator-associated pneumonia (VAP) has been little investigated. In this study we analyzed the incidence of anaerobes in patients with a first episode of bacteriologically documented VAP (> 10(3)CFU/ml), using protected specimen brushes (PSB). We particularly took care to preserve anaerobic conditions during transport and the microbiological procedure. Two groups were considered: group A with anaerobic bacteria recovered from PSB, with or without anaerobes, and group B with aerobic bacteria only. One hundred and thirty patients were included, 30 (23%) in group A, and 100 (77%) in group B. The main anaerobic strains isolated were Prevotella melaninogenica (36%), Fusobacterium nucleatum (17%), and Veillonella parvula (12%). Univariate analysis demonstrated that patients in group A were younger than those in group B (p < 0.05) and their simplified acute physiologic score was higher (p < 0.02). The percentage of patients receiving antibiotics before PSB did not differ significantly between group A (57%) and group B (35%). VAP with anaerobes occurred more often in patients orotracheally intubated than nasotracheally intubated (p < 0.02). Episodes of VAP involving anaerobic bacteria occurred more often in the first five days (early VAP) than after the fifth day (late VAP) (p < 0.05). The 3-mo mortality rate was similar in the two groups, but death occurred earlier in group B (p < 0.01). Multivariate analysis demonstrated that presence of altered level of consciousness (p = 0.0002), higher simplified acute physiologic score (p = 0.003), and admission to the medical ICU (p = 0.02) were the factors independently predisposing to the development of VAP with anaerobes.