We evaluated the hypercapnia response test as a weaning outcome predictor from mechanical ventilation in patients with chronic obstructive pulmonary disease (COPD). We studied 44 COPD intubated patients considered ready for a weaning trial. The hypercapnia test was based on the modified method of re-inhalation of expired air Through the hypercapnic test we calculated the ratio of the change in minute volume (V(E)) to the change in PaCO2 (deltaV(E)/deltaPaCO2), the change in airway occlusion pressure at 0.1 second of inspiration (P0.1) to change in PaCO2 (deltaP0.1/deltaPACO2), the ratio of the change in P0.1 to baseline PaCO2 (deltaP0.1/PaCO2) and the ratio of the change in V(E) to baseline PaCO2 (deltaV(E)/PaCO2). Nineteen patients were successfully weaned and 25 patients failed. No differences in baseline clinical characteristics were found between the two groups. Weaning failure COPD patients had lower deltaP0.1/deltaPaCO2 (0.19 +/- 0.11 and 0.34 +/- 0.20 cm H2O/mmHg respectively, P = 0.006) and lower deltaV(E)/deltaPaCO2 (0.21 +/- 0.15 and 0.40 +/- 0.22 l/min/mmHg respectively, P = 0.002) than successfully weaned patients. The area under the receiver operating characteristic curve to discriminate weaning outcome was for the baseline PaCO2 0.81 (95% confidence interval: 0.66 to 0.91), hypercapnic PaCO2 0.76 (0.61 to 0.88), hypercapnic drive response 0.74 (0.59 to 0.86), hypercapnic ventilatory response 0.76 (0.60 to 0.87), deltaP0.1/PaCO2 0.76 (0.60 to 0.87) and for the deltaV(E)/PaCO2 0.81 (0.67 to 0.91). COPD patients with weaning failure have a significantly more blunted response to the hypercapnia response test than weaning success patients. This test could be useful to predict weaning failure patients if the combined values of the hypercapnic drive and hypercapnic ventilatory response were below the threshold values.