Although pancreatectomy is still performed in a few patients with pancreatic cancer, and nearly all patients who develop pancreatic cancer eventually die of their disease, significant improvements have been made recently. Pancreatectomy is now safer, with major morbidity (hemorrhage, pancreatic anastomotic leak, intraabdominal sepsis) occurring in only about 20% and operative mortality of less than 5%. Two (seemingly subtle) issues cannot be overemphasized when someone carefully studies the literature: (1) There is a crucial difference between actuarial and actual survival, with the former generally being higher whereas the latter is true; and (2) careful re-review of pathologic specimens (especially in long-term survivors) initially diagnosed as pancreatic cancer, preferably by an independent pathologist before publishing long-term results is essential. (Erroneous inclusion of patients with nonductal carcinoma substantially and artificially increases survival.) After curative resection, 5-year actual survival is realistically about 10% with median survivals of 12 to 18 months. In certain subgroups with favorable pathologic characteristics (neoplasms <2 cm without nodal or perineural invasion) the prognosis appears to be significantly better, with the 5-year survival about 20%. The recent improvements in postoperative morbidity and mortality and long-term outcome (resulting also in decreased cost of care of such patients) have occurred typically in centers with an invested interest in and proven record with pancreatic surgery. Further improvements in survival should be sought at the areas of earlier diagnosis and novel treatments designed to prevent locoregional recurrences; the role of extended resections must be determined by prospective, randomized trials.