ObjectiveTo evaluate the feasibility and outcome results of noninvasive mechanical ventilation (NIV) in daily clinical practice outside any prospective protocol-driven trial.DesignAn observational retrospective cohort study.SettingA 22-bed medical intensive care unit in a university hospital.PatientsA consecutive cohort of 124 patients who underwent 143 NIV trials, regardless of the indication, over two consecutive years (1997–1998).InterventionsNone.ResultsA total of 604 acute respiratory failure patients underwent mechanical ventilation, and 143 NIVs were performed in 124 patients. The overall prevalence of NIV use was 143 of 604 patients (24%) in three groups: hypoxemic acute respiratory failure (29.5%), hypercapnic acute respiratory failure (41%), and weaning/postextubation (29.5%). Intubation was avoided in 92 of 143 of the NIVs performed (64%), 19 (13%) after changing the initial NIV mode (i.e., a success rate of 62%, 51%, and 86% in the three groups, respectively). A total of 35 of 51 intubated patients (69%) required intubation during the first 24 hrs of NIV. Intensive care unit stay was 12 ± 10 days for the overall population, and mortality, when NIV failed, was 13 of 124 patients (10.5%). Arterial pH (p= .0527) and the Pao2/Fio2ratio (p= .0482) after 1 hr were the only independent predictive factors for NIV failure by multivariate analysis.ConclusionsThis study confirms the results of controlled trials and demonstrates the feasibility and efficacy of NIV applied in daily clinical practice. These results suggest that NIV should be considered as a first-line ventilatory treatment in various etiologies of acute respiratory failure and as a promising weaning technique and postextubation ventilatory support. However, NIV should certainly be performed by a motivated and sufficiently trained care team.