BACKGROUND: Mobilising patients early after stroke [early mobilisation (EM)] is thought to contribute to the beneficial effects of stroke unit care but it is poorly defined and lacks direct evidence of benefit. OBJECTIVES: We assessed the effectiveness of frequent higher dose very early mobilisation (VEM) after stroke. DESIGN: We conducted a parallel-group, single-blind, prospective randomised controlled trial with blinded end-point assessment using a web-based computer-generated stratified randomisation. SETTING: The trial took place in 56 acute stroke units in five countries. PARTICIPANTS: We included adult patients with a first or recurrent stroke who met physiological inclusion criteria. INTERVENTIONS: Patients received either usual stroke unit care (UC) or UC plus VEM commencing within 24 hours of stroke. MAIN OUTCOME MEASURES: The primary outcome was good recovery [modified Rankin scale (mRS) score of 0-2] 3 months after stroke. Secondary outcomes at 3 months were the mRS, time to achieve walking 50 m, serious adverse events, quality of life (QoL) and costs at 12 months. Tertiary outcomes included a dose-response analysis. DATA SOURCES: Patients, outcome assessors and investigators involved in the trial were blinded to treatment allocation. RESULTS: We recruited 2104 (UK,  = 610; Australasia,  = 1494) patients: 1054 allocated to VEM and 1050 to UC. Intervention protocol targets were achieved. Compared with UC, VEM patients mobilised 4.8 hours [95% confidence interval (CI) 4.1 to 5.7 hours;  < 0.0001] earlier, with an additional three (95% CI 3.0 to 3.5;  < 0.0001) mobilisation sessions per day. Fewer patients in the VEM group ( = 480, 46%) had a favourable outcome than in the UC group ( = 525, 50%) (adjusted odds ratio 0.73, 95% CI 0.59 to 0.90;  = 0.004). Results were consistent between Australasian and UK settings. There were no statistically significant differences in secondary outcomes at 3 months and QoL at 12 months. Dose-response analysis found a consistent pattern of an improved odds of efficacy and safety outcomes in association with increased daily frequency of out-of-bed sessions but a reduced odds with an increased amount of mobilisation (minutes per day). LIMITATIONS: UC clinicians started mobilisation earlier each year altering the context of the trial. Other potential confounding factors included staff patient interaction. CONCLUSIONS: Patients in the VEM group were mobilised earlier and with a higher dose of therapy than those in the UC group, which was already early. This VEM protocol was associated with reduced odds of favourable outcome at 3 months cautioning against very early high-dose mobilisation. At 12 months, health-related QoL was similar regardless of group. Shorter, more frequent mobilisation early after stroke may be associated with a more favourable outcome. FUTURE WORK: These results informed a new trial proposal [A Very Early Rehabilitation Trial - DOSE (AVERT-DOSE)] aiming to determine the optimal frequency and dose of EM. TRIAL REGISTRATION: The trial is registered with the Australian New Zealand Clinical Trials Registry number ACTRN12606000185561, Current Controlled Trials ISRCTN98129255 and ISRCTN98129255. FUNDING: This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in ; Vol. 21, No. 54. See the NIHR Journals Library website for further project information. Funding was also received from the National Health and Medical Research Council Australia, Singapore Health, Chest Heart and Stroke Scotland, Northern Ireland Chest Heart and Stroke, and the Stroke Association. In addition, National Health and Medical Research Council fellowship funding was provided to Julie Bernhardt (1058635), who also received fellowship funding from the Australia Research Council (0991086) and the National Heart Foundation (G04M1571). The Florey Institute of Neuroscience and Mental Health, which hosted the trial, acknowledges the support received from the Victorian Government via the Operational Infrastructure Support Scheme.