Between 1940 and 1978, 179 patients underwent pancreatic resection (64 total, 102 Whipple, 13 distal) at the Presbyterian Hospital, predominantly for carcinoma of the pancreas and periampullary area. With respect to operative morbidity and mortality and survival, these patients have been compared with 141 patients subjected to pancreatic biopsy only, and with 172 by-passed for palliation. Likewise, total pancreatectomy has been compared to pancreaticoduodenectomy (Whipple) in terms of safety and efficacy. The overall major postoperative complication rate for pancreatic resection was 36%, in contrast with 13.5% for biopsy only and 16.8% for by-pass. Of the resected cases with major complications postoperatively, roughly half died, a mortality of 17.9%. Patients who underwent Whipple resections fared significantly better than did those having total pancreatectomies; the postoperative mortality following 102 Whipples was 14.7%, as compared with 23.4% for total pancreatectomies. Intra-abdominal sepsis accounted for most of the postoperative deaths; nine pancreatic and four biliary leaks or fistulae followed Whipple resections. The later complications were of interest; 18 patients undergoing biliary-en-teric by-pass procedures later developed gastroduodenal obstruction, 15 of whom required reoperation, and in 18 survivors of pancreatic resection, upper gastrointestinal hemorrhage (mostly from marginal ulcers) developed, necessitating surgery in seven. Brittle diabetes was a problem in nine patients following pancreatectomy. Survival rates were discouraging in all categories. For ductal carcinoma of the pancreas, median survival for biopsy only was two months, for by-pass six months, for total pancreatectomy nine months, and for Whipple resection 14 months. There were three five-year survivors following resection, a rate of 4.5%. Five-year survival rates following resection for ampullary, common duct, duodenal, and islet cell cancer were 27.8, 33.3, 27.3, and 37.5%, respectively. It is concluded that survival after resection for ductal pancreatic cancer is so rare as to be considered more a biologic aberration than a result of radical surgery. Despite theoretical advantages of total pancreatectomy over Whipple resections, our experience would suggest that the latter can be carried out with lower morbidity and mortality, and with equal chance for cure. Resection for pancreatic cancer should not be abandoned, but rather undertaken with greater selectivity. Operative morbidity and mortality can probably be improved additionally by preoperative transhepatic biliary decompression, and later complications reduced by including vagotomy with gastric resection at the time of pancreatectomy and by performing prophylactic gastroenterostomies in conjunction with by-pass procedures.