Exercise testing was performed on 37 patients with resectable lung lesions who were deemed inoperable because of any of the following risk factors: (1) FEV1 less than or equal to 40 percent of predicted; (2) radionuclide calculated postlobectomy FEV1 less than or equal to 33 percent of predicted; or (3) arterial PCO2 greater than or equal to 45 mm Hg. The patients who reached a peak level of oxygen consumption during exercise (VO2Peak) of greater than or equal to 15 ml/kg/min were offered surgical treatment. Patients with a VO2Peak of less than 15 ml/kg/min were referred for nonsurgical management and excluded from the study. Eight patients underwent lung resection. Their pulmonary function revealed a severe obstructive lung defect with a group mean predicted FEV1 of 40 +/- 6 percent, an FEV1/FVC ratio of 47 +/- 10, a radionuclide calculated postlobectomy FEV1 of 31 +/- 4 percent, and a mean arterial PCO2 of 44 +/- 6 mm Hg. No relationship was found between each patient's exercise performance and spirometric function. Six of the patients had an uncomplicated postoperative course. Two patients had complications but no patient died as a result of surgery or postoperative complications. All patients were discharged from the hospital within 22 days (mean = 9.8 days). We conclude that exercise testing is a useful complement to conventional cardiopulmonary evaluation used in selecting patients for lung resection.