OBJECTIVE We have previously reported our experience with noninvasive positive pressure ventilation (NPPV) via face mask in a small group of selected patients with acute respiratory failure (ARF). NPPV was frequently effective (70% success rate) in correcting gas exchange abnormalities and in avoiding endotracheal intubation (ETI); NPPV also had a low rate of complications. We have evaluated the clinical application of NPPV as first-line intervention in patients with hypercapnic and short-term hypoxemic ARF. A dedicated respiratory therapist conducted an educational program with physicians-in-training rotating through the medical ICUs of a university medical center and supervised implementation of a simplified management protocol. Over 24 months, 164 patients with heterogeneous forms of ARF received NPPV. We report on the effectiveness of NPPV in correcting gas exchange abnormalities, in avoiding ETI, and associated complications, in different conditions precipitating ARF. METHODS One hundred fifty-eight patients completed the study. Forty-one had hypoxemic ARF, 52 had hypercapnic ARF, 22 had hypercapnic acute respiratory insufficiency (ARI), 17 had other forms of ARF, and 26 with advanced illness had ARF and refused intubation. Twenty-five percent of the patients developed ARF after extubation. METHODS Mechanical ventilation was delivered via a face mask. Initial ventilatory settings were continuous positive airway pressure (CPAP) mode, 5 cm H2O, with pressure support ventilation of 10 to 20 cm H2O titrated to achieve a respiratory rate less than 25 breaths/min and an exhaled tidal volume of 7 mL/kg or more. Ventilator settings were adjusted following arterial blood gases (ABG) results. RESULTS The mean duration of NPPV was 25 +/- 24 h. When the 26 patients with advanced illness are excluded, NPPV was effective in improving or correcting gas exchange abnormalities in 105 patients (80%) and avoiding ETI in 86 (65%). Failure to improve ABG values was the reason for ETI in 20 of 46 (43%). The overall average predicted and actual mortality were 32% and 16%, respectively. Survival was 93% in non-intubated patients and 79% in intubated patients. NPPV was effective in lessening dyspnea throughout treatment in all but seven patients. Complications developed in 24 patients (16%). In patients with hypercapnic ARF, nonresponders had a higher PaCO2 at entrance (91.5 +/- 4.2 vs 80 +/- 1.5; p < 0.01). In patients with hypercapnic ARF and ARI, arterial blood gases response (pH and PaCO2) within 2 h of NPPV predicted success (p < 0.0001). None of the entrance parameters predicted need for ETI. CONCLUSIONS We conclude that application of NPPV in clinical practice is an effective and safe alternative to ETI in many hemodynamically stable patients with hypercapnic ARF and in those with hypoxemic ARF in whom the clinical condition can be readily reversed in 48 to 72 h. An educational and supervision program is essential to successfully implement this form of therapy.