© 2015 Elsevier Ltd. Background: Current treatment of ischaemic stroke with thrombolytic therapy is restricted to 3-4·5 h after symptom onset. We aimed to assess the safety and efficacy of desmoteplase, a fibrin-dependent plasminogen activator, given between 3 h and 9 h after symptom onset in patients with occlusion or high-grade stenosis in major cerebral arteries. Methods: In a prospective, double-blind, multicentre, parallel-group, randomised trial, we enrolled patients from 77 hospitals in 17 countries who had ischaemic stroke and occlusion or high-grade stenosis in major cerebral arteries. We randomly assigned patients in a 1:1 ratio, using computer-generated randomisation lists with stratification for baseline National Institutes of Health Stroke Scale and age, to treatment with desmoteplase (90 μg/kg) given 3-9 h after symptom onset or to placebo. Patients, investigators, staff, and the funder were masked to treatment assignment. The primary outcome was a favourable modified Rankin Scale score (0-2) at day 90 in all treated patients who had at least one postbaseline measurement of the modified Rankin Scale. Safety was assessed in all randomly assigned patients who received study drugs. This trial is registered with ClinicalTrials.gov, number NCT00790920. Findings: Between Feb 6, 2009, and Nov 27, 2013, we enrolled 492 patients and randomly assigned 247 to desmoteplase and 245 to placebo (236 in the desmoteplase group and 237 in the placebo group were included in the analysis of the primary endpoint). Median time from stroke onset to treatment was 6·9 h (IQR 5·7-8·0) for placebo and 7·0 h (6·0-7·9) for desmoteplase. Modified Rankin Scale score (0-2) at day 90 occurred in 121 (51%) patients given desmoteplase and 118 (50%) patients given placebo (adjusted odds ratio 1·20, 95% CI 0·79-1·81, p=0·40). 24 (10%) of 240 patients given desmoteplase died compared with 23 (10%) of 238 patients given placebo. Serious adverse events occurred in 64 (27%) of 240 patients receiving desmoteplase compared with 69 (29%) of 238 patients receiving placebo; frequency of symptomatic intracranial haemorrhage (six [3%] patients in the desmoteplase group vs five [2%] in the placebo group), symptomatic cerebral oedema (five [2%] vs four [2%]), and major haemorrhage (ten [4%] vs 15 [6%]) was much the same between treatment groups. Interpretation: Treatment with desmoteplase did not cause safety concerns and did not improve functional outcome when given to patients who had ischaemic stroke and major cerebral artery occlusion beyond 3 h of symptom onset. Funding: H Lundbeck A/S.