ABSTRACT: BACKGROUND:Iron from haemolysed blood is implicated in secondary injury after intracerebral haemorrhage. We aimed to assess the safety of the iron chelator deferoxamine mesylate in patients with intracerebral haemorrhage and to establish whether the drug merits investigation in a phase 3 trial. METHODS:We did a multicentre, futility-design, randomised, placebo-controlled, double-blind, phase 2 trial at 40 hospitals in Canada and the USA. Adults aged 18-80 years with primary, spontaneous, supratentorial intracerebral haemorrhage were randomly assigned (1:1) to receive deferoxamine mesylate (32 mg/kg per day) or placebo (saline) infusions for 3 consecutive days within 24 h of haemorrhage onset. Randomisation was done via a web-based trial-management system centrally in real time, and treatment allocation was concealed from both participants and investigators. The primary outcome was good clinical outcome, which was defined as a modified Rankin Scale score of 0-2 at day 90. We did a futility analysis: if the 90% upper confidence bound of the absolute risk difference between the two groups in the proportion of participants with a good clinical outcome was less than 12% in favour of deferoxamine mesylate, then to move to a phase 3 efficacy trial would be futile. Primary outcome and safety data were analysed in the modified intention-to-treat population, comprising only participants in whom the study infusions were initiated. This trial is registered with ClinicalTrials.gov, number NCT02175225, and is completed. FINDINGS:We recruited 294 participants between Nov 23, 2014, and Nov 10, 2017. The modified intention-to-treat population consisted of 144 patients assigned to the deferoxamine mesylate group and 147 assigned to the placebo group. At day 90, among patients with available data for the primary outcome, 48 (34%) of 140 participants in the deferoxamine mesylate group, and 47 (33%) of 143 patients in the placebo group, had modified Rankin Scale scores of 0-2 (adjusted absolute risk difference 0·6% [90% upper confidence bound 6·8%]). By day 90, 70 serious adverse events were reported in 39 (27%) of 144 patients in the deferoxamine mesylate group, and 78 serious adverse events were reported in 49 (33%) of 147 patients in the placebo group. Ten (7%) participants in the deferoxamine mesylate and 11 (7%) in the placebo group died. None of the deaths were judged to be treatment related. INTERPRETATION:Deferoxamine mesylate was safe. However, the primary result showed that further study of the efficacy of deferoxamine mesylate with anticipation that the drug would significantly improve the chance of good clinical outcome (ie, mRS score of 0-2) at day 90 would be futile. FUNDING:US National Institutes of Health and US National Institute of Neurological Disorders and Stroke.