We assessed the capacity to predict surgical mortality, complications, and functional loss by using the results of resting and exercise respiratory function. Measurements were made before and 4 mo after lung resection in 54 consecutive patients with bronchogenic carcinoma. Predicted postoperative (ppo) FEV1 and DLCO were derived using quantitative lung perfusion scans when baseline FEV1 was < 55% predicted, and by proportional loss of pulmonary segments (total = 19 segments) when FEV1 was > 55% predicted. The patients were aged 67 +/- 7 (mean +/- SD) yr, with an FEV1 of 76 +/- 23% predicted, FEV1/FVC of 55 +/- 13%. and DLCO of 85 +/- 22% predicted. Eleven of the patients had pneumonectomy, 29 had lobectomy, 12 had wedge resection, and two had no resection. Wilcoxon and stepwise logistic regression analyses were used to determine which indices best predicted outcome. Postoperative values were correlated (r = 0.87, p < 0.0001) with actual 4/12 postoperative values of FEV1% and of DLCO (r = 0.56, p < 0.0001). The best predictors (all p < 0.05) for each outcome, in order of usefulness, were as follows. For surgical mortality: (1) the predicted postoperative product (PPP) of ppo FEV1% x ppo DLCO%; (2) ppo DLCO%; (3) ppo FEV1%, and (4) RV, FRC, and SaO2 on the maximal step exercise test. For respiratory complications: body mass index (BMI) (for patients undergoing lobectomy or wedge resection only). For cardiac complications: (1) age; (2) SaO2 at baseline and on the maximal step exercise test; (3) PaO2; (4) PaCO2; and (5) minute ventilation at maximal exercise.(ABSTRACT TRUNCATED AT 250 WORDS)