Between 1946 and 1987, 647 patients with periampullary tumors were diagnosed at the University of Chicago Medical Center. These included 549 tumors located in the head of the pancreas, 40 in the distal common bile duct, 29 in the duodenum, and 29 at the ampulla of Vater. Ninety-eight per cent of all tumors were adenocarcinoma, with 93% of the remaining being duodenal carcinoid or sarcoma. Operability rate ranged from 81% to 97%, according to the tumor location and histologic type. A combination of laparotomy, biopsy, and bypass was performed in 433 patients and only one survived 5 years (0.2%). Resectability rate ranged from 16.5% for pancreatic adenocarcinoma to 89.3% for ampullary tumors. Of the 133 resections, 80 were pancreatoduodenectomies, 29 total pancreatectomies, 7 duodenectomies, 2 gastrectomies, 8 common bile duct resections, and 7 local excisions. Overall 19% of patients who underwent radical resection died in the immediate postoperative period, although mortality has decreased to 5% since 1981. Mortality was 20% after a standard pancreatoduodenectomy and 24.1% after a total pancreatectomy. Five-year actuarial survival rates, including perioperative deaths, were 8.8%, 20%, and 32% for pancreatic, duodenal, and ampullary adenocarcinoma, respectively. One half of patients with sarcoma and two-thirds with carcinoid of the duodenum survived 5 years. No patient with distal common bile duct adenocarcinoma achieved a 5-year survival rate. Multivariate analysis on all patients operated on (n = 566) revealed that the 5-year survival rate was significantly related to intent of operation (palliative 0.2%, curative 12%; p less than 0.001), histologic type (adenocarcinoma 2%, carcinoid and sarcoma 31%; p less than 0.0001), and site (ampullary and duodenal 21%, biliary and pancreatic 0.9%; p less than 0.001). A second multivariate analysis, evaluating only those patients with adenocarcinoma who survived the perioperative period of the radical resection (n = 97) analyzed the influence of tumor size and differentiation, lymphatic, capillary, and perineural microinvasion, lymph node status, and type of procedure (pancreatoduodenectomy vs. total pancreatectomy) on 5-year survival. None of these additional variables was significantly associated with long-term survival rates. In addition we evaluated the presence of local or distant recurrence after resection by analyzing the findings from all autopsies performed on these patients (n = 49): 29.4% of patients died with local recurrence alone, 23.5% with distant recurrence alone, and 47.1% had both local and distant recurrences.(ABSTRACT TRUNCATED AT 400 WORDS)