The ability of preoperative quality-of-life and physiologic variables to predict postoperative complications was tested in 117 consecutive patients undergoing thoracotomy for possible or definite lung cancer. Preoperatively, quality of life was globally assessed by the QLI and Sickness Impact Profile. Dyspnea was assessed by the Clinical Dyspnea Index and a modified Pneumoconiosis Research Unit question. Spirometry and maximal exercise testing were carried out in 115 and 46 subjects, respectively. Thirty-seven percent experienced at least one respiratory complication (eg, pneumonia, atelectasis prompting bronchoscopy, pulmonary embolism). Twofold or greater increases in respiratory complications were associated with current smoking (p < 0.05), cancer as the final pathologic condition (p < 0.10), at least moderate dyspnea (p < 0.10), FEV 1 < 60 percent of predicted (p < 0.05), ventilatory reserve < 25 L (p < 0.05), and VO2max < 1.25 L (p < 0.05). Twofold increases in the incidence of any complication (respiratory, cardiac, etc) were associated with age > or = 75 years (p < 0.05) and cancer as the final pathologic condition (p < 0.05). We conclude that simple historic information (age, smoking status, cancer status, dyspnea) indicates the risk of postoperative morbidity. General quality-of-life measures were not good predictors of morbidity. Our findings corroborate the few studies supporting the value of VO2max and suggest that the usefulness of the ventilatory reserve deserves further attention.