BACKGROUND: Postoperative pancreatic fistula remains the most common complication of pancreaticoduodenectomy (PD) and is potentially lethal. It contributes significantly to prolonged hospitalization and mortality. In this study, we introduced a new technical approach, a modified Roux-en-Y reconstruction and evaluated its safety and feasibility. METHODS: We retrospectively reviewed the patients who had undergone PD with the modified Roux-en-Y reconstructive technique for periampullary malignancies from January 2011 to June 2012. The data on complications, hospital stay and outcomes after the modified Roux-en-Y reconstruction were analyzed. RESULTS: The reconstruction was performed in 171 patients, of whom 92 received pancreaticogastrostomy and 79 received pancreaticojejunostomy. The median duration of surgery was 4.0 hours (range 3.1-6.9) in all patients, and the median blood loss was 530 mL (range 200-2000). Sixty-nine patients were subjected to transfusions, with a median transfusion volume of 430 mL (range 200-1400). The median hospital stay of the patients was 14 days (range 11-38). Their operative mortality was zero and overall morbidity was 18.1% (31 patients). Only four patients (2.3%) developed pancreatic fistulas (grade A fistulas in two patients and grade B in two patients); no patients developed grade C fistula. None of the patients developed bile reflux gastritis. CONCLUSIONS: The modified Roux-en-Y reconstruction, which isolates biliary anastomosis from pancreatic, gastric or jejunal anastomosis, is a safe, reliable, and favorable technique. But it needs further investigation in randomized controlled trials.