There are few data on the use of two-level non-invasive positive pressure ventilation (two-level nIPPV) in the initial treatment of severe acute respiratory failure in emergency departments (ED). In a prospective, non-randomized, pilot study, we assessed (1) the feasability of this method in an ED, (2) its effect on clinical and laboratory data, and (3) its effect on the need of intubation and the final outcome of patients. During a 1-year period all eligible patients admitted for acute respiratory failure, with absence of improvement after periods of specific classic treatments, were included in the study. Each patient received a specific classic treatment and two-level nIPPV with a two-level positive pressure ventilator through a face mask. We recorded parameters on admission, after 15 and 45 minutes of nIPPV and at the end of nIPPV. Sixty-two patients were included: 29 with acute pulmonary oedema (APO), 16 with acute exacerbation of chronic obstructive pulmonary disease (COPD), four with asthma, and 13 with various diseases. In the APO-group, we observed a statistically significant improvement of respiratory and pulse rates, diastolic blood pressure, pH, PaCO2 and SaO2. In acute exacerbation of COPD, we observed only a statistical improvement of respiratory and pulse rates without any significant change of PaCO2 and pH. In the two other groups, there was a clinical, gasometric and haemodynamic improvement in all patients. Four patients were intubated and 10 died, but none in the ED or in the first 24 hours after hospital admission. We were able to institute two-level nIPPV for severe acute respiratory failure in an ED without complications. Its addition to the rest of classic specific treatment seems to bring about a rapid improvement of various clinical and laboratory parameters in most patients. We found no deleterious effect of nIPPV when implemented for short periods of time in the emergency department setting.