Three hundred patients with acute pancreatitis have been studied. Pancreatitis was associated with alcoholism in 207, biliary tract disease in 51 and other conditions in 42. Twenty-two patients died, and an additional 34 patients required more than one week of treatment in the intensive care unit. Retrospective analysis of the first 100 patients identified 11 objective findings which correlated with the occurrence of serious illness or death. They were, on admission, age over 55 years, blood glucose level over 200 milligrams per cent, white blood count over 16,000 per cubic millimeter, serum lactic dehydrogenase level over 350 International units per liter and serum glutamic-oxalacetic transaminase level over 250 Sigma Frankel units per cent. During the initial 48 hours of therapy, the findings were hematocrit value decrease over 10 percentage points, serum calcium level below 8 milligrams per cent, base deficit over 4 milli-equivalents per liter, a blood urea nitrogen level increase over 5 milligrams per cent, estimated fluid sequestration over 6 liters and arterial oxygen tension less than 60 millimeters of mercury. Prospective application of these signs in the latter 200 patients permitted the accurate early identification of those with severe pancreatitis. Only one of 162 patients with fewer than three of these early features was seriously ill or died, while 24 of 38 patients with three or more early positive findings were seriously ill or died. The objective early identification of patients with severe pancreatitis permits more vigorous management of this group and also provides a basis for the selection of patients for the evaluation of proposed improved therapies. Percutaneous peritoneal dialysis in severe pancreatitis was evaluated in ten patients, with three or more positive early signs, who were randomly assigned to dialysis or continued conventional care. Morbidity was strikingly reduced in patients who underwent dialysis, and while death or more than nine days of intensive care occurred in two of five patients who did not receive dialysis, all five patients having dialysis recovered after fewer than nine days of intensive care treatment. Serious illness or death occurred in 31 of the first 100 patients but in only 26 of the more recent 200 patients. There has been a similar fall in mortality from 15.0 to 3.5 per cent. Factors which may contribute to this improvment include the objective early identification of patients with severe disease, the avoidance of early laparotomy whenever practical, the prolongation of nasogastric suction until all evidence of pancreatic inflammation has resolved, careful monitoring of respiratory function and early treatment of pulmonary complications and peritoneal dialysis in patients with severe disease.