A critical analysis of morbidity and mortality for pancreatico-duodenectomy was performed on 221 patients. During the 1960s and 1970s, the morbidity and mortality for pancreaticoduodenectomy were so high that many thought the operative procedure ought to be abandoned. During the 1980s, however, many centers reported mortality rates around 5% and a morbidity of 25% to 35%. Others still reported a mortality of more than 10% and a morbidity of up to 65%. The reasons for these discrepancies are of major interest. In a prospective case-control study 760 patients with malignant and benign diseases of the pancreas were treated in our hospital between September 1, 1985 and April 30, 1997. In 221 cases (128 men, 93 women; mean age 61 years, range 23-83 years) a partial (n = 209) or total (n = 12) pancreaticoduodenectomy, in 12 cases combined with portal vein resection, was performed. Surgical complications were seen in 25%, but less than half of them were severe. General complications were seen in 18.5%. The 30- and 90-day mortality rates were 3.1%, and 5.7%, respectively. In a regression analysis the intraoperative blood loss, preoperative serum bilirubin, diameter of the pancreatic duct, and occurrence of surgical and nonsurgical complications had an independent influence on mortality. In addition to the experience of the surgeon in selecting the patients and his or her personal technical skills when performing a pancreaticoduodenectomy, better anticipation and management of postoperative complications is essential for improving the results of this operation.