Substantial efforts have been devoted to improving the means for early and accurate diagnosis of ventilator-associated (VA) pneumonia in intensive care unit (ICU) patients because of its high incidence and mortality. A good diagnostic yield has been reported from quantitative cultures of bronchoalveolar lavage (BAL) fluid or a protected specimen brush, both obtained by fiberoptic bronchoscopy. As bronchoscopy requires specific skills and is costly, we evaluated a simpler method to obtain BAL fluid, that is, by a catheter introduced blindly into the bronchial tree. Quantitative cultures from bronchoscopically sampled BAL (B-BAL) and blindly nonbronchoscopically collected BAL (NB-BAL) were assessed for sensitivity, specificity, and predictive value for the diagnosis of VA pneumonia. A total of 40 pairs of samples were examined in 28 patients requiring prolonged mechanical ventilation and presenting a high risk of developing pneumonia. For comparison with bacteriologic data we defined a clinical score for pneumonia ranging from zero to 12 using the following variables: body temperature, leukocyte count, volume and character of tracheal secretions, arterial oxygenation, chest X-ray, Gram stain, and culture of tracheal aspirate. To quantify the bacteria in BAL the bacterial index (BI) was used, defined as the sum of the logarithm of the number of bacteria cultured per milliliter of BAL fluid. A good correlation between clinical score and quantitative bacteriology was observed (r = 0.84 for B-BAL and 0.76 for NB-BAL; p < 0.0001). Similar to studies in baboons, patients with pulmonary infection could be distinguished by a BI ⩾ 5 with a sensitivity of 93% and a specificity of 100% (B-BAL). Quantitative culture of blind sampling of BAL resulted in a slightly lower sensitivity (73%) and a specificity of 96% for the diagnosis of pneumonia. When analyzing pairs of B-BAL and NB-BAL samples we found similar results for both qualitative and quantitative bacteriology even if BAL fluids came from different lobes or the contralateral lung. These results suggest that “blind” sampling of BAL can be of value in clinical practice. Microscopic examination of BAL provided rapid (the day of BAL), sensitive (100%), and specific (88%) results, allowing us to introduce early and specific antibiotic therapy.