Several approaches have been used in an attempt to predict the severity and prognosis of attacks of acute pancreatitis. The Ranson and Glasgow criteria include a variety of simple laboratory parameters that are measured on admission and again within 48 h. They are the most widely used indices in clinical practice. The Acute Physiological and Chronic Health Evaluation II system is more complicated, but can be applied to a wide variety of conditions, especially in intensive care settings. The usefulness of this system depends on the threshold score for defining severe pancreatitis; a score of eight appears to be the most appropriate. The finding of nonperfused areas in the pancreas at contrast-enhanced computed tomography is indicative of pancreatic necrosis and portends an unfavourable prognosis. Other clinical and laboratory indices have been proposed, but the most important predictive factor of early mortality seems to be the presence and persistance of a Marshall organ failure score of two or more. This is especially true if organ dysfunction persists beyond 36 h. Radiological findings do not always correlate well with the presence of organ dysfunction, and more investigations are required.