Ventilator-associated pneumonias (VAP) are the most frequent nosocomial lung infections. Histological diagnosis is the standard for reference. A diagnosis of VAP can be considered in presence of fever or hypothermia, hyperleucocytosis or leucopenia, worsening of blood gases, and new radiological infiltrate. Their diagnostical value is dependent on the number of included manifestations. A clinical pulmonary infection score (CPIS) has been produced and compared with histological data. Bacteriological data are essential for an adapted antibiotherapy. The blind non-protected specimen brush is inexpensive and reliable at levels of 10(4) and 10(6) CFU.mL-1, the sensitivity and specificity reach 60%. The double-protected catheter is a sensitive and specific test at the level of 10(3) CFU.mL-1. At present its accuracy has only been compared with bronchial brushing. The culture of a 20 mL mini-broncho-alveolar lavage (same material) is specific (50%) but not sensitive enough (< 70%) at the level of 10(3) CFU.mL-1. The culture of the protected telescopic brush is the most expensive test. It does not carry a risk of contamination, but does not detect a significant amount of VAP even at a level below 10(3) CFU.mL-1. Due to its high specificity it is used as the reference test in numerous studies. The endoscopic broncho-alveolar lavage provides a rapid diagnosis. Although not protected, it carries a low risk of false positives. It also allows the diagnosis of non-bacterial or atypical bacterial lung infections. The diagnosis can also be obtained with lung biopsy which however carries a risk in case of mechanical ventilation, whatever the technique. Except for the protected double catheter, a direct examination has been advocated, for the differentiation between infection and colonization and the improvement of the performances of the simple culture of the broncho-alveolar lavage (search for intra-cellular bacteria).