This retrospective study compares the results of pancreaticogastrostomy (PG) and pancreaticojejunostomy (PJ) in our institution, which has extensive experience in both techniques. Between the years of June 1995 and June 2001, 214 patients underwent pancreaticoduodenectomy (PD) at our institution. Of these 177 had PG and 97 had pancreatojejunostomy (PJ). There were 117 (54.6%) males and 97 (45.3%) females with a mean age of 64.2 +/- 12.4 years. Indications for surgery were pancreatic adenocarcinoma in 101 (47.2%), ampullary adenocarcinoma in 36 (16.9%), distal bile duct adenocarcinoma in 22 (10.2%), duodenal adenocarcinoma in 9 (4.2%), and miscellaneous causes in 46 (21.4%) of patients. Preoperatively, significant differences in the groups were that the patients undergoing PJ were significantly younger than those undergoing PG. Also noted preoperatively, was that the patients undergoing PG had a significantly lower direct bilirubin than those undergoing PJ. With regard to intraoperative parameters, operative time was significantly shorter in the PJ group when compared to the PG group. When the patients who did not develop fistula (N = 186) were compared to those who developed fistula (N = 28) the significant differences were that the patients who developed fistula were more likely to have hypertension preoperatively and a higher alkaline phosphatase. They also showed a significantly higher drain amylase and were likely to have surgery for ampullary, distal bile duct or duodenal carcinoma rather than pancreatic adenocarcinoma. In addition, those patients who developed fistula had a significantly longer postoperative stay, a larger number of intraabdominal abscesses and leaks at the biliary anastomosis. Thirty-day mortality was significantly higher in the PJ group compared to the PG (4 vs. 0, P = 0.041). There was a significantly larger number of bile leaks in the PJ group when compared to the PG (6 vs. 1, P = 0.048). In addition, the PJ group required a significantly larger number of new CT guided drains to control infection (8 vs. 2, P = 0.046) and the PJ group required a larger number of re-explorations to control infection or bleeding (5 vs. 0, P = 0.018). However, the pancreatic fistula rate was not different between the two groups (12% [PG] vs. 14% [PJ]). This retrospective analysis shows that safety of PG can be performed safely and is associated with less complications than PJ and proposes PG as a suitable and safe alternative to PJ for the management of the pancreatic remnant following PD.