To evaluate the correlation between predicted postoperative lung function and postoperative respiratory morbidity, 156 patients with lung cancer who underwent resection were classified into four groups based on the degree of postoperative problems: Group 1--no problems (116 patients); Group 2--retention of sputum or atelectasis requiring bronchofiberscopy two or more times (17 patients); Group 3--tracheostomy or mechanical ventilation for more than 2 days or both (14 patients); and Group 4--postoperative death (9 patients). The mean ages of Groups 2, 3, and 4 were significantly (p less than 0.05) higher than the mean age of Group 1. The predicted postoperative lung function (F) was assessed by the formula F = [1-(b-n)/(42-n)] x f, where f is the preoperative vital capacity or forced expiratory volume in one second, b is the number of subsegments of the resected lung lobe, and n is the number of subsegments obstructed by the tumor, which was assessed by the findings on the chest tomogram, on the bronchogram, at bronchofiberscopy, or a combination of these. The total number of subsegments was assumed to be 42. The predicted postoperative % FEV1 was 65.1 +/- 19.3% in Group 1,55.3 +/- 10.6% in Group 2,37.6 +/- 12.1% in Group 3, and 42.3 +/- 18.4% in Group 4. It was significantly (p less than 0.05) different between all the groups except between Groups 3 and 4. All 10 patients with a predicted postoperative % FEV1 of less than 30% were in Groups 3 and 4. We conclude that special attention to postoperative management is needed for patients whose predicted postoperative %FEV1 is lower than 30%.