Laparoscopic pancreatoduodenectomy may improve postoperative recovery compared with open pancreatoduodenectomy. However, there are concerns that the extensive learning curve of this complex procedure could increase the risk of complications. We aimed to assess whether laparoscopic pancreatoduodenectomy could reduce time to functional recovery compared with open pancreatoduodenectomy.This multicentre, patient-blinded, parallel-group, randomised controlled phase 2/3 trial was performed in four centres in the Netherlands that each do 20 or more pancreatoduodenectomies annually; surgeons had to have completed a dedicated training programme for laparoscopic pancreatoduodenectomy and have done 20 or more laparoscopic pancreatoduodenectomies before trial participation. Patients with a benign, premalignant, or malignant indication for pancreatoduodenectomy, without signs of vascular involvement, were randomly assigned (1:1) to undergo either laparoscopic or open pancreatoduodenectomy using a central web-based system. Randomisation was stratified for annual case volume and preoperative estimated risk of pancreatic fistula. Patients were blinded to treatment allocation. Analysis was done according to the intention-to-treat principle. The main objective of the phase 2 part of the trial was to assess the safety of laparoscopic pancreatoduodenectomy (complications and mortality), and the primary outcome of phase 3 was time to functional recovery in days, defined as all of the following: adequate pain control with only oral analgesia; independent mobility; ability to maintain more than 50% of the daily required caloric intake; no need for intravenous fluid administration; and no signs of infection (temperature <38·5°C). This trial is registered with Trialregister.nl, number NTR5689.Between May 13 and Dec 20, 2016, 42 patients were randomised in the phase 2 part of the trial. Two patients did not receive surgery and were excluded from analyses in accordance with the study protocol. Three (15%) of 20 patients died within 90 days after laparoscopic pancreatoduodenectomy, compared with none of 20 patients after open pancreatoduodenectomy. Based on safety data from the phase 2 part of the trial, the data and safety monitoring board and protocol committee agreed to proceed with phase 3. Between Jan 31 and Nov 14, 2017, 63 additional patients were randomised in phase 3 of the trial. Four patients did not receive surgery and were excluded from analyses in accordance with the study protocol. After randomisation of 105 patients (combining patients from both phase 2 and phase 3), of whom 99 underwent surgery, the trial was prematurely terminated by the data and safety monitoring board because of a difference in 90-day complication-related mortality (five [10%] of 50 patients in the laparoscopic pancreatoduodenectomy group vs one [2%] of 49 in the open pancreatoduodenectomy group; risk ratio [RR] 4·90 [95% CI 0·59-40·44]; p=0·20). Median time to functional recovery was 10 days (95% CI 5-15) after laparoscopic pancreatoduodenectomy versus 8 days (95% CI 7-9) after open pancreatoduodenectomy (log-rank p=0·80). Clavien-Dindo grade III or higher complications (25 [50%] of 50 patients after laparoscopic pancreatoduodenectomy vs 19 [39%] of 49 after open pancreatoduodenectomy; RR 1·29 [95% CI 0·82-2·02]; p=0·26) and grade B/C postoperative pancreatic fistulas (14 [28%] vs 12 [24%]; RR 1·14 [95% CI 0·59-2·22]; p=0·69) were comparable between groups.Although not statistically significant, laparoscopic pancreatoduodenectomy was associated with more complication-related deaths than was open pancreatoduodenectomy, and there was no difference between groups in time to functional recovery. These safety concerns were unexpected and worrisome, especially in the setting of trained surgeons working in centres performing 20 or more pancreatoduodenectomies annually. Experience, learning curve, and annual volume might have influenced the outcomes; future research should focus on these issues.Grant for investigator-initiated studies by Johnson & Johnson Medical Limited.