Fiberoptic bronchoscopy (FOB) may worsen oxygenation and clinical status in severely hypoxemic patients. We conducted a prospective, randomized double-blind trial to compare the delivery of continuous positive airway pressure (CPAP) as a tool for maintaining oxygenation during FOB, to the delivery of oxygen only. Thirty consecutive patients who needed FOB for diagnostic purposes were enrolled. Their arterial oxygen pressure (Pa(O(2))) to inspired oxygen fraction (FI(O(2))) ratio was below 300 mm Hg. CPAP was generated by a simple new device open to the atmosphere. During FOB and the 30 min thereafter, pulse oximetry values (Sp(O(2))) were significantly higher in the CPAP than the Oxygen group (95.7 +/- 1.9% versus 92.6 +/- 3.1, p = 0.02). The lowest Sp(O(2)) values were observed in the Oxygen group (93.5 +/- 2.4% versus 88.6 +/- 3.4, p = 0.002). Arterial blood gases 15 min after FOB showed that Pa(O(2)) had increased in the CPAP group and decreased in the Oxygen group (DeltaPa(O(2)) = +10.5% +/- 16.9 versus -15% +/- 16.6, p = 0.01). Five patients in the Oxygen group, but none in the CPAP group, developed respiratory failure in the 6 h after FOB and required ventilatory assistance (p = 0.03). We conclude that in hypoxemic patients, the use of a new CPAP device during FOB allowed minimal alterations in gas exchange and prevented subsequent respiratory failure.