Ventilator-associated pneumonia (VAP) is one of the most common causes of morbidity and mortality in intensive care unit patients. However, the diagnosis is quite difficult. Gram stain (GS) of bronchoalveolar lavage (BAL) sample is a time-saving diagnostic method for VAP. However, its clinical significance has not been adequately investigated. The aim of this study was to determine its sensitivity and specificity for VAP diagnosis. We prospectively performed GS and quantitative bacterial cultures (QBC) of BAL samples, obtained through fiberoptic bronchoscope, in 75 consecutive postoperative and/or multiple trauma patients with suspected VAP. We considered BAL-GS as positive for VAP diagnosis when (i) polymorphonuclear neutrophils were > 25 per optic field at a magnification x 100 (p.o.f x 100); (ii) squamous epithelial cells were < 1% p.o.f. x 100; and (iii) one or more microorganisms were seen p.o.f., at a magnification x 1000 (p.o.f., x 1000). VAP was diagnosed with criteria similar to those used in previous studies. Pneumonia was the final diagnosis in 22/75 (29%) patients. The BAL-GS was positive in 17/22 patients with VAP and in 7/53 patients without VAP. Accordingly, the sensitivity of BAL-GS for VAP diagnosis was 77%, the spccificity 87%, the positive predictive value 71% and the negative predictive value 90%. Our data suggest that BAL-GS has good sensitivity and high specificity for VAP diagnosis. It could therefore constitute a useful complementary tool in the task of early diagnosis and treatment of VAP.