BACKGROUND The objective was to identify predictive factors for non-invasive ventilation (NIV) failure and to describe its use in bronchiolitis. METHODS Prospective observational study that included patients diagnosed with bronchiolitis with a modified Wood's Clinical Asthma Score 5, or oxygen saturation <92%, or venous CO(2) partial pressure (PCO(2)) 60 mm Hg, with no response to medical treatment, who received NIV from December 2005 to May 2008. We collected clinical data before NIV began and at 1, 6, 12, 24 and 48 h. Need for intubation was considered as NIV failure. RESULTS NIV was successful in 83% of 47 cases included. Patients in whom NIV failed had lower weight (5.2+/-2.2 vs. 3.5+/-0.8 kg, P=.011), lower age [1.8 (0.3-12.3) vs. 0.8 (0.4-4.3) months, P=.038)], lower heart rate (HR) before NIV began (176.3+/-19.1 vs. 160.4+/-9.7 beats/minute, P=.010), lower HR decrease at hours 1 (-16.0+/-17.3 vs.+1.1+/-11.6, P=.005) and 12 (-31.5+/-19.7 vs. -0.75+/-12.2, P=.002), presence of apnoeas (23.1% vs. 75%; P=.004) and of a predisposing condition (84.6% vs. 50%; P=.029). Multivariate analysis identified the absence of a predisposing condition, and a greater HR decrease during the first hour as success-associated independent factors (OR 0.004; 95% CI 0.000-0.664 and OR 0.896; 95% CI: 0.809-0.993, respectively). CONCLUSIONS NIV has a high success rate in bronchiolitis. The main parameters which can predict NIV success are the absence of a predisposing condition and a higher HR decrease in the first hour.