The results of treatment and outcome in 52 consecutive patients presenting to Leicester from 1972 to 1984 are presented. The number of patients diagnosed increased from two per year before the introduction of duodenoscopy to nearly five per year afterwards. Endoscopic drainage (ED) was attempted in 21 patients with a success rate of 81 per cent. In eight cases ED was used pre-operatively and in the remainder as definitive treatment. Twenty-four patients had a Whipple's resection (12.5 per cent mortality), four patients had a local resection (no deaths), ten patients had surgical bypass (60 per cent mortality) and thirteen patients had ED alone (23 per cent mortality). The major risk factor score was significantly greater in patients undergoing surgical bypass compared with Whipple's resection. Age and risk factor scores were significantly greater in patients who had ED drainage alone than in surgical patients. The 5 year survival rate for resection was 56 per cent versus 13 per cent for drainage procedures (P less than 0.001). Survival in resection cases was directly related to the degree of tumour differentiation and a new staging system. It is proposed that all patients with ampullary tumours should have endoscopic biopsy followed by ED; Whipple's resection remains the surgical treatment of choice.