Background Early enteral nutrition is crucial for preventing malnutrition and improving outcomes in patients with severe stroke, but previous trials have provided conflicting results regarding the optimal nutritional strategy. We aimed to compare the efficacy and safety of three enteral feeding strategies in patients with severe stroke. Methods The Optimizing Early Enteral Nutrition in Severe Stroke (OPENS) study was a multicentre, investigator initiated, prospective, open-label, randomised controlled trial, with blinded outcome assessment, in 16 tertiary and district general hospitals in the west of China. Adult patients with acute severe ischaemic or haemorrhagic stroke (Glasgow Coma Scale score <= 12 or National Institutes of Health Stroke Scale score >= 11 on admission) who were expected to receive enteral nutrition for more than 7 days were randomly assigned (1:1:1) to full enteral nutrition (70-100% of estimated caloric requirements), modified full enteral nutrition (full enteral nutrition plus prokinetic agents), or hypocaloric enteral nutrition (40-60% of estimated caloric requirements) via a centralised web-based randomisation system. The assigned nutrition was initiated within 24 h after enrolment and continued for 7 days. The computer-generated randomisation sequence was prepared by a statistician not involved with the rest of the study. Randomisation was done with an automated permuted block size of six. The allocation was unblinded to participants and investigators. The primary efficacy outcome was the proportion of participants with poor outcome (modified Rankin Scale score >= 3) at day 90 and the prespecified primary safety outcome was mortality at day 90, assessed in the intention-to-treat population. The trial is registered with ClinicalTrials.gov, NCT02982668. Findings Between Jan 15, 2017, and Sept 23, 2020, 321 patients were randomly assigned (107 in each group) and 315 patients (175 [56%] men, median age 71 years, IQR 60-78) were included in the final analysis. The study was terminated ahead of schedule on Sept 23, 2020, because a significant difference between groups was detected in mortality. The proportion of participants with poor outcomes at 90 days did not differ (modified full enteral nutrition 86 [82%] of 105 patients vs full enteral nutrition 85 [80%] of 106 patients, adjusted odds ratio [OR] 0middot87, 95% CI 0middot41-1middot86, p=0middot721; hypocaloric enteral nutrition 76 [73%] of 104 patients vs full enteral nutrition 0middot61, 0middot30-1middot27, p=0middot186; hypocaloric enteral nutrition vs modified full enteral nutrition 0middot70, 0middot34-1middot46, p=0middot340). Hypocaloric enteral nutrition showed significantly higher 90-day mortality than did modified full enteral nutrition (35 [34%] of 104 patients vs 18 [17%] of 105 patients, adjusted OR 2middot89, 95% CI 1middot46-5middot72; p=0middot0023), whereas the difference was not significant between hypocaloric enteral nutrition and full enteral nutrition (24 [23%] of 106 patients; adjusted OR 1middot92, 95% CI 1middot00-3middot69; p=0middot049), and between modified full enteral nutrition and full enteral nutrition (adjusted OR 0middot61, 0middot29-1middot28; p=0middot187). The most common adverse event was pneumonia, the incidence of which showed no significant difference among groups (full enteral nutrition 82 [78%] of 105 patients, modified full enteral nutrition 83 [81%] of 103 patients, hypocaloric enteral nutrition 78 [75%] of 104 patients; p=0middot625). Interpretation In the early phase of severe stroke, modified full enteral nutrition or hypocaloric enteral nutrition did not significantly reduce the risk of a poor outcomes compared with full enteral nutrition over a 90-day period. Hypocaloric enteral nutrition might be associated with increased mortality compared with modified full enteral nutrition. Further studies are needed to investigate whether modified full enteral nutrition might be the optimal strategy. Copyright (c) 2022 Elsevier Ltd. All rights reserved.