To determine whether intra-operative videoendoscopy provides useful information on the integrity of gastrointestinal anastomoses, laboratory and clinical studies were undertaken. In the dog model, the videoendoscope was inserted per orum to visualize side-to-side (gastrojejunostomy) and end-to-end (jejunojejunostomy) anastomoses and to establish endoscopic criteria of normal anastomosis, verified by examination of the specimen. Technically faulty anastomoses were then constructed and inspected endoscopically to determine the signs of poor technique. In clinical application, the videoendoscope was used to inspect the anastomoses after low anterior resection, revision of gastric bypass, and choledochoduodenostomy. The laboratory results indicate that direct visualization of a defect in suture-line with the end-viewing endoscope is difficult, although indirect signs of trouble are helpful. Conversely, the appearance of a perfect anastomosis without any indirect signs of poor technique is accurate in the assurance of a trouble-free anastomosis. The best clinical use is in low anterior resection of the rectum and the worst is in choledochoduodenostomy. It is concluded that videoendoscopic inspection of anastomosis is only of limited help in determining the integrity of a difficult anastomosis, and much work is required to perfect both the instrument and technique before general use is recommended.