INTRODUCTION: The aim is to describe weaning procedures, weaning failure rates, and predictors and consequences of weaning failure in infants admitted to pediatric intensive care units (PICUs) for severe bronchiolitis. METHODS: This is a multicenter prospective observational cohort study in five PICUs in French university hospitals. Consecutive infants aged 3 days to 6 months admitted between November 2020 and April 2022 with a clinical diagnosis of severe bronchiolitis requiring noninvasive ventilatory support by bilevel positive airway pressure (BiPAP), continuous positive airway pressure (CPAP), or high-flow nasal cannula (HFNC). RESULTS: Demographic and clinical data were collected prospectively. Weaning strategies were classified as direct, HFNC for de-escalation, and gradual with decreasing support levels. Multivariate analysis was performed to identify independent predictors of weaning failure. Of the 135 included patients (median age 1 [1-2] months), 60 (44%), 49 (36%), and 26 (19%) were managed by HFNC-based, direct, and gradual weaning, respectively. Bronchiolitis severity was similar in the three groups. By multivariate analysis, predictors of weaning failure was gradual weaning (odds ratio, 10.56 [2.87-38.86], p <
0.01), while apnea at admission (0.26 [0.07-0.96], p = 0.04) and younger age (0.44 [0.23-0.84], p = 0.02) were protective factors. PICU length of stay was shorter with HFNC-based weaning (3.8 [1.9-5.4] days vs. 4.3 [3.0-6.9] and 5.1 [3.8-7.4] with direct and gradual weaning, respectively, p = 0.02). CONCLUSIONS: Among patients with severe bronchiolitis, a weaning strategy using HFNC for de-escalation was associated with shorter PICU stays. Whether this method also decreases the risk of weaning failure deserves investigation.