For many years, the classical Whipple procedure has been the standard operative treatment for resectable periampullary cancer. Recently, the pylorus preservation operation has also been applied to periampullary tumours. This newer procedure has potential advantages in terms of ease of performance and postoperative physiology, but has yet to be proved as an adequate cancer operation. We, therefore, undertook the present retrospective analysis to compare the outcome following the pylorus preservation operation (n = 13) with that of the classical Whipple resection (n = 13) in 26 patients with histologically proved adenocarcinoma of the head of the pancreas. The two groups of patients were comparable for age, sex, pre-operative laboratory data, and resected margins free from tumour. More patients undergoing the classical Whipple resection had Stage I disease (10 versus 4, P less than 0.05). However, this difference would not be significant if the two patients in the pylorus preservation group with carcinomas in situ were considered to have stage I disease. Mean tumour diameter in the pylorus preservation patient group (3.2 +/- 0.6 cm) was smaller (P less than 0.05) than in the classical Whipple group (4.1 +/- 1.0 cm) but more patients in the pylorus preservation group also had metastases to the regional lymph nodes (54 versus 23 per cent). Overall operative morbidity (31 per cent) and mortality (4 per cent) was acceptable and did not differ between the two groups. Five year actuarial survival for pylorus preservation (25 per cent) was comparable to that observed for the Whipple procedure in this and other series.