The majority of lethal complications after pancreatic head resection are due to septic complications after leakage from the pancreatojejunostomy. Especially the smooth pancreatic remnant is prone to develop parenchymal leaks from shear forces applied during tying of the sutures. We developed a new mattress technique that avoids such shear forces, and we compared this method to the standard Cattell (duct-to-mucosa) technique. A total of 113 patients undergoing standard pancreatic head resection were prospectively randomized to receive either the standard Cattell anastomosis (n = 56) or the new mattress technique (n = 57). All patients were evaluated for surgical and medical complications until discharge. Primary diagnosis and further demographic data compared well between the groups. The time to perform the mattress anastomosis was significantly shorter (15 vs. 22 minutes; p < 0.0001). The incidence of complications at the pancreatojejunostomy, and the length of hospital stay and survival were not significantly different between the two groups; however, a trend toward more reoperations was noted in the Cattell group (10 vs. 5; p < 0.097). The new mattress technique is simple, and our data show that the two techniques yield similar incidences of complications. Therefore the mattress technique for pancreatojejunostomy seems to be safe and is, in our opinion, well suitable for training schedules in pancreatic surgery.