The MR CLEAN-LATE trial provided evidence for the safety and efficacy of endovascular treatment for acute ischaemic stroke within the late window (after 6-24 h) in patients who were preselected based on the presence of collateral flow on CT angiography. We aimed to evaluate clinical outcomes 2 years after randomisation. MR CLEAN-LATE was a phase 3, multicentre, open-label, blinded-endpoint, randomised controlled trial conducted at 18 stroke intervention centres in the Netherlands. If endovascular treatment could be initiated within 6-24 h of symptom onset or last seen well, patients (aged 18 years or older) with an acute ischaemic stroke due to a large vessel occlusion in the anterior circulation and at least some collateral flow in the affected middle cerebral artery territory on CT angiography were randomly assigned (1:1) to either endovascular treatment with best medical treatment (endovascular treatment group) or best medical treatment alone (control group). Web-based randomisation, stratified by centre, was performed with the use of permuted blocks (block size eight to 20). The researchers who collected clinical outcomes and analysed the results were masked to treatment allocation; treating physicians, local investigators, and patients were aware of the received treatment. The primary outcome of MR CLEAN-LATE was the modified Rankin Scale (mRS) score at 90 days after randomisation. For this 2-year prespecified analysis, the primary outcome was mRS score at 2 years (minus 3 months to plus 6 months). Primary and safety analyses were performed based on the modified intention-to-treat principle, and included patients who provided (deferred) consent or died before consent could be obtained. Missing data were handled with multiple imputation by chained equations. The trial is completed and is registered at ISRCTN, ISRCTN19922220. Between Feb 2, 2018, and Jan 27, 2022, 535 patients were randomly assigned in the MR CLEAN-LATE trial, of whom 502 (94%) gave deferred consent and comprised the modified intention-to-treat population (255 in the endovascular treatment group and 247 in the control group). 261 (52%) patients were female and 241 (48%) were male. Data for mRS score at 2 years were available for 226 (89%) patients in the endovascular treatment group and for 202 (82%) patients in the control group. The median mRS score at 2 years was 4 (IQR 2-6) in the endovascular treatment group and 6 (2-6) in the control group. The endovascular treatment group demonstrated a shift towards better functional outcomes on the mRS (adjusted common odds ratio 1·41 [95% CI 1·00-1·99]; p=0·049). All-cause mortality at 2 years was 34% (87 of 255) in the endovascular treatment group and 41% (101 of 247) in the control group (adjusted hazard ratio 0·81 [95% CI 0·60-1·08]; p=0·15). Major vascular events (ie, transient ischaemic attack, ischaemic stroke, haemorrhagic stroke, and cardiac events) were reported between 90 days and 2 years in 23 patients in the endovascular treatment group and 13 patients in the control group. Our results show that the effectiveness of late-window (after 6-24 h) endovascular treatment in improving clinical outcomes is sustained for up to 2 years in a population preselected based on the presence of collateral flow on CT angiography. This finding might be important for prompting further evaluations of cost-effectiveness, health-care policy development, and clinical decision making. The Dutch Organization for Health Research and Health Innovation (ZonMW), Collaboration for New Treatments of Acute Stroke Consortium, Dutch Heart Foundation, Stryker, Medtronic, Cerenovus, Health Holland Top Sector Life Sciences & Health, and the Netherlands Brain Foundation.