It is estimated that by 2050, there will be a 300% increase in the elderly population (> or =65 years) and a corresponding increase in elderly patients presenting for surgical evaluation. Surgical decision-making in this population can be difficult because outcomes in the elderly are poorly defined. We reviewed 2698 consecutive pancreaticoduodenectomies (PDs) at our institution over a 35-year period (April 1970 through March 2005), with the last 1000 resections being done in the last 4 years. Data collected included surgical indication, mortality (defined as 30-day or in-hospital mortality), complications, and survival. Patients were divided by age into three groups (<80, 80-89, and > or =90 years) and evaluated using multiple logistic regression. Two hundred seven patients > or =80 years old underwent a PD (7.7% of 2698). Patients 80-89 years of age had a mortality rate of 4.1% (8 of 197) and a complication rate of 52.8% (99 of 197), whereas patients < or =79 years of age had a mortality of 1.7% and a complication rate of 41.6% (P < 0.05). There were no perioperative deaths among the 10 patients > or =90 years of age, and their complication rate was 50% (5 of 10). One-year survival for patients 80-89 years of age was 59.1%, and that for patients > or =90 years was 60%. Age was not an independent risk factor for perioperative mortality and morbidity following PD after adjusting for preoperative comorbidities. We demonstrate that PD can be safely performed in patients over 80 years of age and conclude that age alone should not be a contraindication to pancreatic resection. The advent of improved surgical outcomes and an aging population will likely result in a significant increase in the number of PDs performed in the next few decades.