BackgroundAlthough high-flow nasal cannula therapy (HFNC) has become a popular mode of non-invasive respiratory support (NRS) in critically ill children, there are no randomised controlled trials (RCTs) comparing it with continuous positive airway pressure (CPAP). We performed a pilot RCT to explore the feasibility, and inform the design and conduct, of a future large pragmatic RCT comparing HFNC and CPAP in paediatric critical care.MethodsIn this multi-centre pilot RCT, eligible patients were recruited to either Group A (step-up NRS) or Group B (step-down NRS). Participants were randomised (1:1) using sealed opaque envelopes to either CPAP or HFNC as their first-line mode of NRS. Consent was sought after randomisation in emergency situations. The primary study outcomes were related to feasibility (number of eligible patients in each group, proportion of eligible patients randomised, consent rate, and measures of adherence to study algorithms). Data were collected on safety and a range of patient outcomes in order to inform the choice of a primary outcome measure for the future RCT.ResultsOverall, 121/254 eligible patients (47.6%) were randomised (Group A 60%, Group B 44.2%) over a 10-month period (recruitment rate for Group A, 1 patient/site/month; Group B, 2.8 patients/site/month). In Group A, consent was obtained in 29/33 parents/guardians approached (87.9%), while in Group B 84/118 consented (71.2%). Intention-to-treat analysis included 113 patients (HFNC 59, CPAP 54). Most reported adverse events were mild/moderate (HFNC 8/59, CPAP 9/54). More patients switched treatment from HFNC to CPAP (Group A: 7/16, 44%; Group B: 9/43, 21%) than from CPAP to HFNC (Group A: 3/13, 23%; Group B: 5/41, 12%). Intubation occurred within 72 h in 15/59 (25.4%) of HFNC patients and 10/54 (18.5%) of CPAP patients (p = 0.38). HFNC patients experienced fewer ventilator-free days at day 28 (Group A: 19.6 vs. 23.5; Group B: 21.8 vs. 22.2).ConclusionsOur pilot trial confirms that, following minor changes to consent procedures and treatment algorithms, it is feasible to conduct a large national RCT of non-invasive respiratory support in the paediatric critical care setting in both step-up and step-down NRS patients.Trial registrationclinicaltrials.gov, NCT02612415. Registered on 23 November 2015.Electronic supplementary materialThe online version of this article (10.1186/s13054-018-2080-3) contains supplementary material, which is available to authorized users.