Lung resection in patients with cardiopulmonary dysfunction is associated with increased risk. We studied 52 elderly male patients with airflow obstruction and a lung mass. Studies were performed at rest with routine ventilatory tests and lung scan quantitation of right-left lung function. Cycle ergometry exercise was then performed at 2 submaximal work loads (25 and 40 watts). Data were obtained using systemic and pulmonary artery catheterization for blood pressures, thermal dilution cardiac output, and blood gases. Twenty-nine patients underwent lung resection and seven failed to tolerate the procedure (death within 60 days or prolonged ventilator dependence). Those parameters most clearly separating the group tolerating surgery (n = 22) from the intolerant group (n = 7) were obtained during exercise and included: cardiac index (tolerant 5.5 +/- 1.3 vs intolerant 3.9 +/- 0.3 L/min/m2, p less than .01), O2 delivery (p less than .01) and calculated VO2 ml/kg/min (tolerant 11.3 +/- 2.1 vs intolerant 7.8 +/- 1.5 ml/kg/min, p less than .001). Pulmonary vascular pressures and calculated resistance did not predict intolerance. Calculated VO2 at 40 watts did not separate those patients who had survivable complications from those who did not (p much greater than .05). Multivariate analysis suggests that exercise VO2 is an important predictor of tolerance of lung resection because it reflects the effects of cardiac function and O2 transport. In our patients with COPD, submaximal exercise testing predicted intolerance of lung resection better than calculation using quantitative lung scanning. Exercise testing may accomplish this goal by uncovering deficits in O2 transport.