Background/Purpose: Pancreatic leakage is a leading cause of morbidity and mortality after pancreatico-duodenectomy (PD). Pancreaticogastrostomy (PG) has been reported to be associated with a lower pancreatic leakage rate and morbidity rate than pancreaticojejunostomy (PI). This study compared the preoperative characteristics, surgical risk factors, intraoperative parameters, and postoperative outcome between PJ and PG. Methods: From March 1992 to March 2005, a comparative study between PI and PG for patients with periampullary lesions undergoing PD was conducted. A total of 377 consecutive patients underwent PD. Among them, 188 patients under-went PJ and 189 underwent PG. Results: The overall mortality, morbidity and pancreatic leakage following PD were 5%, 45.1% and 10.6%, respectively. The mortality, morbidity and pancreatic leakage were 8.9%, 56.4% and 17.6% in the PJ group, and 2.1%, 33.9% and 3.7% in the PG group (p < 0.001). Mean operative time was 9.3 hours versus 6.7 hours (p < 0.001), mean blood loss was 1032 mL versus 891 mL (p = 0.064) and mean hospital stay was 34.8 days versus 26.1 days (p < 0.001) in the PJ and PG groups, respectively. PJ, soft pancreas, pancreatic duct stenting, low surgical volume (< 20) and age (> 65 years) were identified as risk factors for pancreatic leakage, while PJ, soft pancreas, pancreatic duct stenting and low surgical volume (< 20) were four significant risk factors for surgical morbidity. Further, PJ, pancreatic leakage, low surgical volume (< 20) and age (> 65 years) were identified to be surgical risk factors for mortality. Conclusion: PG is a safer method than PI following PD as a significantly lower rate of pancreatic leakage, surgical morbidity and mortality, shorter operation time, and shorter postoperative hospital stay are reported.