Pancreatic cancer has been extensively researched in recent years, but overall survival after diagnosis is almost unchanged since the time of Whipple. In the meantime, we have tried to determine the factors influencing surgical risk, resectability and survival. Between 1968 and 1992, 516 patients with pancreatic cancer were monitored; 160 patients with cancer of the body-tail were excluded. Eighty-five of the remaining 356 patients with pancreatic head cancer were resected, while the remainder underwent only palliative procedures. Surgical outcome, in terms of operative mortality or complications, was correctly predicted preoperatively in > 80% of patients. The preoperative evaluation of the resectability of pancreatic cancer has been investigated with different, mainly invasive, procedures. CT scan, associated with angiography and laparoscopy is reported to give better results, with a resectability rate up to 78%. 67% of our patients who were diagnosed as resectable according to CT-scan features and serum CA 19-9 < 200 U/ml, were actually resected; furthermore 40% had a potentially curative resection. Most of our 'curative' resections (80%) were within this group of patients. Multivariate analysis showed only 4 factors influencing long-term survival: TNM stage, diabetes, age > 70 years, tumour grading.