To evaluate dynamic contrast enhanced computed tomography for detecting and localising pancreatic necrosis in acute pancreatitis.|Prospective evaluation with blind reporting of scans.|Single teaching hospital.|60 Consecutive patients with acute pancreatitis suspected to have pancreatic necrosis because of major organ system failure (13); slow recovery five to seven days after admission with raised scores on the acute physiological and chronic health evaluation (APACHE-II) system (27); or findings on previous ultrasonography or computed tomography (20).|Pancreatic necrosis proved histologically--that is, greater than 30 g necrotic tissue debrided at laparotomy (for life threatening sepsis or peritonitis) or necropsy.|Dynamic computed tomography correctly localised pancreatic necrosis in 11 patients (confirmed at laparotomy in nine and at necropsy in two). Of nine patients with low enhancement of peripancreatic tissues alone, eight recovered after conservative management; necropsy confirmed viable pancreas and necrosis of peripancreatic fat in one patient. Of 40 patients with normal contrast enhancement, none required laparotomy to debride pancreatic necrosis.|Dynamic contrast enhanced computed tomography seems to be a safe and accurate method of identifying and localising pancreatic and peripancreatic necrosis, which cannot be predicted accurately by conventional imaging techniques.