Between 1984 and 1987, 472 patients with histologically or cytologically verified carcinomas of the pancreas or papilla of Vater, were accrued in the Norwegian Pancreatic Cancer Trial. Surgical assessment revealed resectability in 29% (94 of 330) of the pancreatic tumours and 89% (25 of 28) of the papillar tumours. Tumours of the pancreatic head were resectable in 32% (84 of 259). The sensitivities of the different diagnostic methods in patients with resectable tumours were: FNAC (fine needle aspiration cytology) 80%, ERCP (endoscopic retrograde cholangio-pancreatography) 78%, PTC (percutaneous transhepatic cholangiography) 73%, ERCP with duct cytology 67%, CT (computed tomography) 58%, US (ultrasound) 42% and angiography 22%. The positive predictive values (PV+) in resectable disease were: US 29%, CT 35%, ERCP 43% and angiography 44%. Corresponding figures for unresectable disease were US 95%, CT 97%, ERCP 75% and angiography 88%. Resectable tumours of the pancreas and papilla of Vater had an average macroscopic diameter of 3.2 x 3.4 cm and 2.2 x 2.3 cm, respectively. Tumour size increased with stage. Increasing tumour size and abdominal pain combined with short diagnostic delay both decreased resectability rate, whereas a combination of long diagnostic delay and abdominal pain had a more favorable resectability rate. Radical pancreatic surgery, if effective in the treatment of carcinoma of the pancreas or papilla of Vater, should not be undertaken if any preoperative diagnostic test demonstrates signs of indisputable unresectability. Available methods for the evaluation of resectability in patients lacking such signs are insufficient. This necessitates exploratory laparotomy in many patients.