BACKGROUND: Although pneumonia is the third most common type of nosocomial infection following urinary tract infection and surgical wound infection, it is associated with the highest mortality rate (28-37%), and 16% of deaths in hospitals are caused by it. The purpose of this study was to analyze our cases of postoperative pneumonia and to establish the principles of prevention and treatment for postoperative pneumonia. METHODS: Thirteen cases diagnosed as postoperative pneumonia out of 11,227 patients who were operated on from Jan. 1994 to June 1997 at the Department of Surgery, Seoul National University Hospital, were analyzed by a retrospective review of their medical records. The diagnosis of postoperative pneumonia was based on the `Center for Disease Control' criteria. RESULTS: The incidence of postoperative pneumonia was 0.12%. The median patient age was 58 years (ranging from 31 to 70 years). There were 7 males and 6 females. The causal diseases for the operations were 7 gastric cancers, 2 ileus, and 4 others. Five subtotal gastrectomies, 3 total gastrectomies, and 5 other operations were performed, and there were 10 elective and 3 emergent operations. The preoperative status was investigated. Two cases were preoperatively diagnosed as bronchiectasis. Other associated medical illnesses were 2 coronary arterial diseases, 2 hypertensions, and 2 others. Six patients had a history of smoking. The `American Society of Anesthesiologist' physical status score was checked in 11 cases. There were 2 cases with one points, six cases with two points, and 3 cases with a higher number of points. The average time until diagnosis after operation was 4.0 +/- 3.85 days. All had purulent sputum and high fever, and showed pneumonic infiltration on Chest PA. Four Gram positive and 7 Gram negative bacteria were isolated from the sputum of 12 patient, including three cases with Pseudomonas aeruginosa and two cases with MRSA. All patients were treated with antibiotics. Five patients were admitted to the ICU, and 3 patients maintained intubation after anesthesia. Twelve patients had a nasogastric tube, for which the average period of intubation was 14.1 +/- 14.61 days. One patient expired, and 12 improved. CONCLUSIONS: There was a very low incidence of postoperative pneumonia, and most patients were old. Upper abdominal surgery proved to be an important risk factor, and prolonged maintenance of the nasogastric tube may predispose postoperative patients to pneumonia. The appearance of many antibiotics- resistant bacteria demanded the judicious use of antibiotics.