The incidence of complications after gastrectomy increases signicantly when extended lymph node dissection is added [1,2]. However, the reported incidence of complications varies considerably in di!erent settings, and these di!erences can prove instructive. There are few prospective randomised studies comparing di!erent methods of managing postoperative complications in this group of patients, in whom there are specic risks, such as pancreatic stula. It is therefore necessary to concede that the superiority of one approach to managing complications cannot be considered scientically proven. Nevertheless, surgeons remain interested in the results of institutions with large experience of this kind of surgery, because it seems likely that they would have evolved e!ective methods for dealing with and preventing the complications. This is especially the case where the institutions can show a low incidence of complications, and a low incidence of death and other serious consequences in their patients who do develop complications. This article is based on the collected experience of Japanese surgeons at the National Cancer Center Hospital in Tokyo. The advice given on the management of complications is based on this experience, and not on randomised trial evidence. Where the practice advocated di!ers from that in the reader's institution, it may be appropriate to consider an RCT as a method of evaluating the best approach, provided this is considered ethical. Tables 1 and 2 show the frequency of specic complications after gastrectomy observed in the Dutch Gastric Cancer Trial (DGCT) and in the consecutive case series at the National Cancer Center Hospital, Tokyo (NCCH) in the 1980s [3], respectively. In both of these tables, the incidence of anastomotic leakage is high. From the observations at NCCH, however, it is evident that the incidence of anastomotic leakage has decreased remarkably during the period of study. This decline correlates temporally with the increasing use of circular staplers (staple guns) for oesophago-jejunostomy. Table 3 shows the leakage results associated with the learning curve for the technique of using this instrument, although this is here partially confounded by improvements in the design of the instrument itself. The recent incidence of leakage at the oesophago-jejunostomy site in our institution is less than 1% and most of these occur in cases with an intra-thoracic oesophago-jejunostomy. Between 1993 and 1997, 513 patients underwent esophagojejunostomy. Only seven (1.4%) had anastomotic leakage, and of these four had a handsewn anastomosis and ve had an intra-thoracic anastomosis. The risk of leakage at the stapled esophagojejunostomy was 0.6%. In the DGCT, the incidence of anastomotic leakage after distal gastrectomy was signicantly higher following gastro-duodenostomy (Billroth 1: 18.8%) than after gastrojejunostomy (Billroth 2 or Roux-en-Y: 6.9%). This relative risk corresponds well with the experience at NCCH, although the absolute risk of leakage at either type of anastomosis was much lower in the Japanese series: In 1993, leakage was seen in seven out of 181 patients after Billroth 1 at NCCH, and none of 53 following Roux-en-Y reconstruction. Since 1993, there has been no occurrence of leakage at the gastro-jejunostomy at NCCH. Similar ndings on relative risk suggest that Billroth type1 (B1) reconstruction should be avoided after D2 distal gastrectomy. The major di!erences in absolute risk re#ect some of the many di!erences between the series: specialist surgery by teams with vast experience of the operation in younger, leaner patients with less co-morbid pathology in Japan, versus surgery by more generalist surgeons during a learning phase in older, more obese patients with considerable cardiorespiratory co-morbidity in the Netherlands. The most frequent major complication after a D2 or more extended lymphadenectomy is pancreatic juice leakage and/or left subphrenic abscess. The incidence of this complication after a total gastrectomy is highest when the distal pancreas is resected together with spleen,