To assess the usefulness of a method for predicting postoperative peak oxygen uptake based on lost lung function after lung resection (VO2peak-PPO) and to establish the underlying physiological foundation for the relation between VO2peak-PPO and the measured postoperative value VO2peak-PO), we studied 29 patients (26 men) [age 60 (SD9)] with chronic airflow limitation [FEV1 = 66 (SD13)%] undergoing lobectomy or major pulmonary resection to treat lung cancer. The patients were assigned to groups according to whether postoperative exercise tolerance was considered to be limited by exhaustion of ventilatory reserve (LV) or not (NLV). Data to estimate postoperative pulmonary function was obtained one week before surgery: patients performed pulmonary function tests and exercise tests on a treadmill; dyspnea was also evaluated and perfusion scintigraphs were obtained. Pulmonary function, exercise tolerance and dyspnea were evaluated again approximately five months after surgery. The mean difference between VO2peak-PPO and VO2peak-PO was -0.034 (CI 0.293 to -0.348) l.min-1 and the between-group correlation coefficient was 0.76. The correlation between VO2peak-PPO and VO2peak-PO was 0.86 (SE 0.1) [0.89 (SE 0.13) for LV (n = 14) patients and 0.85 (SE 0.16) for NLV (n = 15) patients]. The correlations after adjusting for preoperative VO2peak-PPO were 0.73 (SE 0.2) and 0.35 (SE 0.27) for LV and NLV patients, respectively. We conclude that VO2peak-PPO provides a valid but only moderately precise estimate of VO2peak-PO. Only in LV patients is there a true relation between a decrease in VO2peak and loss of lung function.