This retrospective analytic study evaluated whether abnormal diffusing capacity for carbon monoxide (DLCO) is a predictor of postoperative morbidity and mortality in patients undergoing pneumonectomy for lung cancer. The medical records of patients undergoing pneumonectomy at Vancouver General Hospital between January 1992 and December 1997 were reviewed. Postoperative complications occurring within 30 days of resection were classified into mortality, and cardiovascular, pulmonary, and technical complications. A total of 151 pneumonectomy cases were reviewed. There were 100 men (66%) and 51 women (34%) with a mean age of 61 years. Complications occurred in 73 patients (48%), including mortality in eight (5%), cardiovascular morbidity in 50 (33%), pulmonary morbidity in 30 (20%), and technical morbidity in 22 (15%). Arrhythmia (21%) and pulmonary edema (13%) were the two major cardiovascular complications. Patients with complications had a greater smoking history, a longer hospital stay, a lower forced expiratory volume in 1 second (FEV1), a lower FEV1/forced vital capacity (FVC) ratio, a lower DLCO, and a lower DLCO/alveolar volume (VA) ratio than patients without complications. A DLCO of 70% predicted was the best functional predictor of postoperative complications, with a complication rate of 94% in patients with a DLCO of less than 70% predicted compared with 27% in patients with a DLCO of at least 70% predicted (sensitivity, 62%; specificity, 96%). However, technical morbidity was not related to preoperative lung function variables, including DLCO. Patients with a DLCO of at least 70% predicted had a low postpneumonectomy complication rate. Although cardiac arrhythmia was the major cause of morbidity, pulmonary edema was the major cause of mortality.