We compared noninvasive positive-pressure ventilation (NPPV), using bilevel positive airway pressure, with usual medical care (UMC) in the therapy of patients with acute respiratory failure (ARF) in a prospective, randomized trial. Patients were subgrouped according to the disease leading to ARF (chronic obstructive pulmonary disease [COPD], a non-COPD-related pulmonary process, neuromuscular disease, and status postextubation), and were then randomized to NPPV or UMC. Thirty-two patients were evaluated in the NPPV group and 29 in the UMC group. The rate of endotracheal intubation (ETI) was significantly lower in the NPPV than in the UMC group (6.38 intubations versus 21.25 intubations per 100 ICU days, p = 0.002). Mortality rates in the intensive care unit (ICU) were similar for the two treatment groups (2.39 deaths versus 4.27 deaths per 100 ICU days, p = 0.21, NPPV versus UMC, respectively). Patients with hypoxemic ARF in the NPPV group had a significantly lower ETI rate than those in the UMC group (7.46 intubations versus 22.64 intubations per 100 ICU days, p = 0.026); a similar trend was noted for patients with hypercapnic ARF (5.41 intubations versus 18.52 intubations per 100 ICU days, p = 0.064, NPPV versus UMC, respectively). Patients with ARF in the non-COPD category had a lower rate of ETI with NPPV than with UMC (8.45 intubations versus 30.30 intubations per 100 ICU days, p = 0.01). Although the rate of ETI was lower among COPD patients receiving NPPV, this trend did not reach statistical significance (5.26 intubations versus 15.63 intubations per 100 ICU days, p = 0.12, NPPV versus UMC, respectively). In conclusion, NPPV with bilevel positive airway pressure reduces the rate of ETI in patients with ARF of various etiologies.