BACKGROUND: Pilot clinical trials have shown the safety of intra-arterial bone marrow mononuclear cells (BMMNCs) in stroke. However, the efficacy of different doses of intra-arterial BMMNCs in patients with acute stroke has not been tested in a randomised clinical trial. We aimed to show safety and efficacy of two different doses of autologous intra-arterial BMMNC transplantation in patients with acute stroke. METHODS: The IBIS trial was a multicentre phase 2, randomised, controlled, investigator-initiated, assessor-blinded, clinical trial, in four stroke centres in Spain. We included patients (aged 18-80 years) with a non-lacunar, middle cerebral artery ischaemic stroke within 1-7 days from stroke onset and with a National Institutes of Health Stroke Scale score of 6-20. We randomly assigned patients (2:1:1) with a computer-generated randomisation sequence to standard of care (control group) or intra-arterial injection of autologous BMMNCs at one of two different doses (2 × 106 BMMNCs/kg or 5 × 106 BMMNCs/kg). The primary efficacy outcome was the proportion of patients with modified Rankin Scale scores of 0-2 at 180 days in the intention-to-treat population, comparing each BMMNC dose group and the pooled BMMNC group versus the control group. The primary safety endpoint was the proportion of serious adverse events. This trial was registered at ClinicalTrials.gov, NCT02178657 and is completed. FINDINGS: Between April 1, 2015, and May 20, 2021, we assessed 114 patients for eligibility. We randomly assigned 77 (68%) patients: 38 (49%) to the control group, 20 (26%) to the low-dose BMMNC group, and 19 (25%) the high-dose BMMNC group. The mean age of participants was 62·4 years (SD 12·7), 46 (60%) were men, 31 (40%) were women, all were White, and 63 (82%) received thrombectomy. The median NIHSS score before randomisation was 12 (IQR 9-15), with intra-arterial BMMNC injection done a median of 6 days (4-7) after stroke onset. The primary efficacy outcome occurred in 14 (39%) patients in the control group versus ten (50%) in the low-dose group (adjusted odds ratio 2·08 [95% CI 0·55-7·85]; p=0·28), eight (44%) in the high-dose group (1·89 [0·52-6·96]; p=0·33), and 18 (47%) in the pooled BMMNC group (2·22 [0·72-6·85]; p=0·16). We found no differences in the proportion of patients who had adverse events or dose-related events, but two patients had a groin haematoma after cell injection in the low-dose BMMNC group. INTERPRETATION: Intra-arterial BMMNCs were safe in patients with acute ischaemic stroke, but we found no significant improvement at 180 days on the mRS. Further clinical trials are warranted to investigate whether improvements might be possible at different timepoints. FUNDING: Instituto de Salud Carlos III co-funded by the European Regional Development Fund/European Social Fund, Mutua Madrileña, and the Regional Ministry of Health of Andalusia.