Background There is no consensus on the best method of preventing postoperative pancreatic fistula (POPF) after pancreaticoduodenectomy (PD). This multicentre, parallel group, randomized equivalence trial investigated the effect of two ways of pancreatic stenting after PD on the rate of POPF. Methods Patients undergoing elective PD or pylorus‐preserving PD with duct‐to‐mucosa pancreaticojejunostomy were enrolled from four tertiary referral hospitals. Randomization was stratified according to surgeon with a 1 : 1 allocation ratio to avoid any related technical factors. The primary endpoint was clinically relevant POPF rate. Secondary endpoints were nutritional index, remnant pancreatic volume, long‐term complications and quality of life 2 years after PD. Results A total of 328 patients were randomized to the external (164 patients) or internal (164) stent group between August 2010 and January 2014. The rates of clinically relevant POPF were 24·4 per cent in the external and 18·9 per cent in the internal stent group (risk difference 5·5 per cent). As the 90 per cent confidence interval (−2·0 to 13·0 per cent) did not fall within the predefined equivalence limits (−10 to 10 per cent), the clinically relevant POPF rates in the two groups were not equivalent. Similar results were observed for patients with soft pancreatic texture and high fistula risk score. Other postoperative outcomes were comparable between the two groups. Five stent‐related complications occurred in the external stent group. Multivariable analysis revealed that soft pancreatic texture, non‐pancreatic disease and high body mass index (23·3 kg/m2 or above) predicted clinically relevant POPF. Conclusion External stenting after PD was associated with a higher rate of clinically relevant POPF than internal stenting. Registration number: NCT01023594 (https://www.clinicaltrials.gov). Internal stenting is better