From 1981 to 1990, 14 of 70 patients hospitalized at our institution for severe acute pancreatitis were selected to undergo percutaneous drainage of pancreatic abscess, under computed tomographic (CT) scan guidance. Pancreatic abscess was defined, on contrast-enhanced CT scan, as an infected fluid collection without pancreatic necrosis. There were nine men and five women, ranging in age from 28 to 46 years. The main cause of pancreatitis was alcohol abuse (eight patients). Other causes were gallstones (two patients), hyperlipidemia (two patients), postoperative (one patient) and one unknown. Ranson criteria were available in ten patients and ranged from three to six. Percutaneous drainage was performed as the primary treatment in 13 patients and for removal of a residual collection postoperatively in one patient. In two critically ill patients, percutaneous drainage was performed as a temporizing measure. In 12 patients with well-limited hypodense collections, percutaneous drainage was expected to result in the definitive cure of the abscess. Pigtail drains (No. 14F), were inserted using local anesthesia and CT scan guidance. Two patients had two drains and 12 patients had only one drain. Two patients were definitively cured by percutaneous drainage and all other patients were operated upon for removal of infected necrosis. In this study, the lack of accuracy of contrast-enhanced CT scan in the diagnosis of peripancreatic necrosis is highlighted and that percutaneous drainage has a better efficiency in the treatment of residual collections postoperatively than as a primary treatment of infected fluid collections is illustrated.