Pulmonary infection is thought to be a common complication of ARDS. We undertook this prospective study to determine the incidence of pulmonary infection in patients with ARDS, and to evaluate the impact of nosocomial pneumonia on severity of ARDS and on survival. Two hundred one bronchoscopies were performed in 105 patients with ARDS with retrieval of distal airway secretions by bronchoalveolar lavage (BAL) and protected specimen brush (PSB). Whenever possible, bronchoscopy was performed at predetermined times: Day 3, Day 7, Day 14, and Day 21 after the onset of ARDS. The majority of patients were receiving antibiotics at the time of study. Changes in bacterial flora over time were determined by quantitative cultures of BAL and PSB. Bacterial growth was common, but usually at small concentrations. Only 16 patients met quantitative culture criteria for pneumonia (PSB > or = 10(3) cfu/ml or BAL > or = 10(4) cfu/ml). Correlation was poor between clinical evidence of pneumonia and pneumonia by quantitative culture criteria: clinical criteria had a very low sensitivity (24%) for predicting positive quantitative culture results, and a low specificity (77%) for predicting negative quantitative culture results. There was no correlation between total colony counts on BAL or PSB and severity of ARDS as judged by Pao2/FIo2 ratios, days receiving ventilation, or compliance. Furthermore, there was no correlation between bacterial growth and survival. We conclude that pneumonia defined by quantitative bacteriology is uncommon in ARDS. The potentially confounding role of broad-spectrum antibiotics should be studied further.