BackgroundEngland has high morbidity and mortality from house fires, with steep social gradients for child deaths: 106 people were admitted to hospital in 2012–13, and 44 deaths between 2008 and 2012 were in children under 5 years old. We developed an evidence-based fire-prevention intervention for use in children's centres comprising an injury prevention briefing (IPB), training, and facilitation for IPB implementation. MethodsWe conducted a multicentre cluster-randomised controlled trial, with a nested qualitative study, in four English study sites. Parents who were at least 16 years old with a child younger than 3 years were recruited from children's centres in disadvantaged areas. Once 30 parent participants were recruited at each centre, centres were stratified by trial site, randomly allocated within strata to one of three arms by an independent statistician using randomisation algorithm in Stata with permuted block randomisation and block size of three: IPB plus facilitation (IPB+), IPB only, and usual care (control). Training (month 0) and facilitation were provided for centres allocated IPB+ at months 1, 3, and 8. Both intervention arms completed follow-up interviews at 12 months. The primary outcome was the proportion of families reporting a fire-escape plan. 11 centres per arm were required to detect an absolute difference of 20% or more in families with fire-escape plans in either intervention arm compared with control (80% power, 5% significance level, intraclass correlation (ICC)=0·05), assuming responses from 20 families per centre. Treatment arms were compared by use of multilevel models to account for clustering by centre. Ethics approval was given by Derbyshire ethics committee (11/EM/0011). Written informed consent was obtained from individual parents and children's centres. This trial is registered with ClinicalTrials.gov, NCT01452191. Findings1112 parents at 36 centres were recruited (373 IPB+, 369 IPB only, 370 control). At 12 months 751 (68%) parents (241 IPB+, 252 IPB only, 258 control) were followed up. Possession of a fire-escape plan was reported by 120 (51%) IPB+ participants, 108 (44%) IPB only, and 119 (47%) controls. There was no significant effect on family possession of fire-escape plans (IPB+ vs control odds ratio 1·41, 95% CI 0·91–2·20; IPB only vs control 0·93, 0·58–1·49 [ICC=0·003]). No adverse events were reported. InterpretationThe IPB, training, and facilitation did not increase parents' possession of fire-escape plans. Both the difficulties faced by children's centres in delivering the intervention to participating families and the secondary outcomes help to explain this outcome. Further research is required to explore both enhancement of penetration of interventions, such as the IPB, to disadvantaged communities and what single item questions about fire escape plans actually measure. FundingNational Institute for Health Research under its Programme Grants for Applied Research programme (RP-PG-0407-10231).