The prognosis of carcinoma of the pancreas remains dismal, with no recent appreciable improvement in overall survival. Prognostication in individual patients has become important in the selection of appropriate treatment, particularly in advanced disease with the option of non-operative stenting. Patients with a short life expectancy can be identified by a poor performance status (e.g. Karnofsky) or when metastatic disease is detected by imaging (ultrasonography or CT) or laparoscopy. The correlation between tumour staging and prognosis remains problematic. The anatomical location of the pancreas and lack of representative tissue sampling invariably result in understaging, which limits the value of tumour staging systems and hampers the interpretation of therapeutic trials. While it is generally agreed that pancreatic resection provides the only hope for cure, there is no consensus on how radical surgery should be. No convincing data have been produced to support radical (regional) pancreatectomy as a better cancer operation. Improved 5-year survival figures have been achieved after pancreatectomy irrespective of the extent of the resection. Lower operative mortality has contributed to these improved results.