Fever is one of the most common reasons for consultation in the paediatric emergency department (ED). Because of fear of bacterial infection in parents and caregivers, clinicians often overprescribe laboratory tests and empirical antibiotic treatment. The aims of this study are to demonstrate that using a procalcitonin (PCT) rapid test-based prediction rule (1) would not be inferior to usual practice in terms of morbidity and mortality (non-inferiority objective) and (2) would result in a significant reduction in antibiotic use (superiority objective). This prospective multicentric cluster-randomised study aims to include 7245 febrile children aged 6 days to 3 years with a diagnosis of fever without source in 26 participating EDs in France and Switzerland during a 24-month period. During first period, all children will receive usual care. In a second period, a point-of-care PCT-based algorithm will be used in half of the clusters. The primary endpoints collected on day 15 after ED consultation will be a composite outcome of death or intensive care unit admission for any reason, disease-specific complications, diagnosis of bacterial infection after discharge from the ED for the non-inferiority objective and proportion of children with antibiotic treatment administered for the superiority objective. The endpoints will be compared between the two groups (experimental and control) by using a mixed logistic regression model adjusted on clustering of participants within centres and period within centres. If the algorithm is validated, a new strategy will be discussed with medical societies to safely manage fever in young children without the need for invasive procedures for microbiological testing or empirical antibiotics. This study was submitted to an independent ethics committee on 17 May 2018 (no. 2018-A00252-53). Results will be submitted to international peer-reviewed journals and presented at international conferences. NCT03607162; Pre-results. © Author(s) (or their employer(s)) 2020. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.