To refine the functional guidelines for operability for lung resection, we prospectively studied 55 consecutive patients with suspected lung malignancy thought to be surgically resectable. Lung function and exercise capacity were measured preoperatively and at 3 and 12 months postoperatively. Preoperative pulmonary scintigraphy was used to calculate the contribution to overall function by the affected lung or lobe and to predict postoperative lung function. Pneumonectomy was performed in 18 patients, lobectomy in 29, and thoracotomy without resection in six. No surgery was attempted in two patients who were considered functionally inoperable. Cardiopulmonary complications developed in 16 patients within 30 days of surgery, including three deaths. The predictions of postoperative function correlated well with the measured values at 3 months. For FEV1, r = 0.51 in pneumonectomy (p less than 0.05) and 0.89 in lobectomy (p less than 0.001). Predicted postoperative FEV1 (FEV1-ppo), diffusing capacity (DLCO), predicted postoperative DLCO (DLCO-ppo) and exercise-induced arterial O2 desaturation (delta SaO2) were predictive of postoperative complications including death and respiratory failure. In patients who underwent pneumonectomy, the best predictor of death was FEV1-ppo. The predictions were enhanced by expressing the value as a percentage of the predicted normal value (% pred) rather than in absolute units. For the entire surgical group a FEV1-ppo greater than or equal to 40% pred was associated with no postoperative mortality (n = 47), whereas a value less than 40% pred was associated with a 50% mortality (n = 6), suggesting that resection is feasible when FEV1-ppo is greater than or equal to 40% pred.(ABSTRACT TRUNCATED AT 250 WORDS)