Bogaard et al10 compared CO measured by means of an impedance cardiography device (IPG-104 impedance; Mini-Lab; Detroit, MI) and by means of the noninvasive CO2 rebreathing method. SV and CO were measured at rest and during steady-state exercises, ranging from light intensity to exhaustion, in patients with moderate COPD.,10 The authors found a good correlation between the two methods during exercise, with few data falling outside the limits of agreement of ± 22%. The mean CO difference (impedance − rebreathing) was 0.01 ± 1.28 L/min, which are acceptable values. It must be underlined that these results were obtained in COPD patients with lesser obstruction than our subjects. Bogaard et al10 observed higher impedance CO values compared to rebreathing at low-intensity exercise and lower impedance CO values at high-intensity exercise. It is interesting to note that the slope of the regression analysis of CO vs V̇o2 with their impedance cardiography device was 6.2 L/L; this is much less than what we observed in our study and close to the slope we reported for COfick. Bogaard et al,,33 in a review, hypothesized that different parameters could contribute to the inaccuracy of an electrical impedance cardiography device in COPD patients: increased motion artifacts induced by accessory respiratory muscle movement, increased respiratory artifacts (higher ventilation per minute for the same quantity of work), and change in lung volumes.33All these factors could be responsible for the discrepancies we observed between COfick and COpf, but it must be underlined that we observed no correlation between COpf − COfick and minute ventilation, tidal volume, or breathing frequency. We found no more correlation between COpf − COfick and hyperinflation or obstruction estimated at rest in our study. This is in agreement with the results reported for healthy subjects breathing with a high resistance.34 Nevertheless, we cannot exclude hyperinflation as a factor explaining the inaccuracy of the COpf measurements in our study. Indeed, Bogaard et al,33 explained that baseline Z, which accounts for the basic thoracic Z, compensates for lung volumes. With our device, baseline Z is not measured, thus preventing perhaps such compensation. However, during the calibration, maximum Z − minimum Z, which is probably influenced by lung volumes, is taken into account; this minimizes probably the fact that baseline Z is not measured. However, CO measurement is based on the aortic component of the dZ/dt estimation. During exercise in COPD patients, nonnegligible changes in pulmonary vascular flow and intrathoracic pressure, in addition to hyperinflation, may change the passage of the current emitted by the impedance cardiography device through the chest and thus modify the aortic component of dZ/dt. This may explain that adequacy was not improved during exercise, in contrast to what was observed by Bogaard et al10 in their patients with less severe COPD. Moreover, in their study, Critchley and Critchley35 observed that CO measured by impedance cardiography underestimated the real CO in patients with increased amount of fluid in the chest. They suggested that the aortic component of dZ/dt was minored by the pulmonary component. It is not excluded that in patients with increased amount of air in their chest (in COPD hyperinflated patients), the opposite situation occurs, which could lead to an overestimation of CO.