Low exercise capacity is considered predictive for postoperative complications or death after thoracic and general surgery. However, in recent literature no agreement has been found about the predictive cut-off values for preoperative exercise parameters. The aim of this work was to investigate whether peak oxygen consumption (V'o2) and noninvasive anaerobic threshold (AT) determined by gas exchange threshold (GET) can be reliable preoperative predictors of mortality and morbidity after lung resection in patients with mild-to-moderate (forced expiratory volume in one second (FEV1) > 50% predicted) chronic obstructive pulmonary disease (COPD). Fifty tour COPD patients were studied before lung surgery: 12 had severe complications, 16 had mild and 26 had no complications. Peak V'O2 sensitivity and specificity in predicting severe postoperative complications were 41.6% and 95.5% respectively (using 75% of the predicted value as cut-off), while for GET they were 91.6% and 97.6% respectively (using 14.5 mL.kg-1.min-1 as cut-off value). Only one patient (3.5%) with a peak V'O2 > 20 mL.kg-1.min-1 suffered severe complications. On the other hand 11 out of the 26 patients (42.3%) with peak V'O2 < 20 mL.kg-1.min-1 had serve complications. In patients with peak V'O2 < 20 mL.kg-1.min-1, 11 out of 12 (91.6%) with a GET < or = 14.5 mL.kg-1.min-1 suffered severe complications, whereas 15 out of 15 (100%) with a GET > 14.5 mL.kg-1.min-1 showed no or mild complications. In conclusion, peak oxygen consumption values > 20 mL.kg-1.min-1 can be considered a safe upper cut-off limit for pulmonary resection. In patients with a peak oxygen consumption value < 20 mL.kg-1.min-1, gas exchange threshold determination can improve significantly the predictivity of a cardiopulmonary test for severe complications and must be routinely considered.