Although Helicobacter pylori (H. pylori) seropositivity is linked to an excess risk of pancreatic cancer, the biologic mechanism is unknown. Gastric ulcer is primarily associated with corpus colonization of H. pylori, atrophic gastritis and formation of N-nitrosamines. Duodenal ulcer is a marker of antral colonization, hyperacidity and uninhibited secretin release. We estimated relative risks for pancreatic cancer among patients with gastric or duodenal ulcer, based on a register-based retrospective cohort study with 88,338 patients hospitalized for gastric ulcer and 70,516 patients for duodenal ulcer recorded in the Swedish Inpatient Register between 1965 and 2003. Following operation, the 14,887 patients who underwent gastric resection and 8,205 with vagotomy were analyzed separately. Multiple record-linkages allowed complete follow-up and identification of all incident cases of pancreatic cancer until December 31, 2003. Standardized incidence ratios (SIRs) estimated relative risks. During years 3-38 of follow-up, we observed a 20% excess risk (95% confidence interval [CI] 10-40%) for pancreatic cancer among unoperated gastric ulcer patients. The excess increased to 50% (95% CI 10-110%) 15 years after first hospitalization (p for trend 5 0.03). SIR was 2.1 (95% CI 1.4-3.1) 20 years after gastric resection. Unoperated duodenal ulcer was not associated with pancreatic cancer risk, nor was vagotomy. Our results lend indirect support to the nitrosamine hypothesis, but not to the hyperacidity hypothesis in the etiology of pancreatic cancer. ' 2006 Wiley-Liss, Inc.Key words: pancreatic cancer; gastric ulcer; duodenal ulcer; gastrectomy; vagotomy; risk; cohort Pancreatic cancer is the most deadly neoplasm in humans.1 Little is known about its etiology apart from known relationships with smoking, a family history, and some rare genetic susceptibility syndromes. Alleged causal links with diabetes and nonhereditary chronic pancreatitis remain under debate.2 Possible association between a history of gastroduodenal diseases and pancreatic cancer was first suggested by an early case-control study, 3 in which a strong association between subtotal gastrectomy and pancreatic cancer was reported. The authors postulated that one of the possible mechanisms is via increased Nnitroso compounds in gastric juice after the operation. Recently, a moderately strong association with Helicobacter pylori (H. pylori) infection was reported. 4,5 Although confounding from unknown causal factors cannot be ruled out as an explanation, there are at least three hypothetical models for the influence of H. pylori on pancreatic carcinogenesis. Firstly, bacterial invasion of the organ and local chronic inflammation, is conceivable but there are no reports in the literature of H. pylori growth in pancreatic tissue or ducts. The second, more plausible, hypothesis involves predominantly gastric antral colonization, hyperchlorhydria, uninhibited secretin release, attendant elevations in basal pancreatic bicarbonate output and pancreatic ductular hyperp...