Introduction: Impedance cardiography (IC) derived from morphological analysis of the thoracic impedance signal is now commonly used for noninvasive assessment of cardiac output (CO) at rest and during exercise. However, in Chronic Obstructive Pulmonary Disease (COPD), conflicting findings put its accuracy into question. Objectives: We therefore compared concurrent CO measurements captured by IC (PhysioFlow: CO IC ) and by the indocyanine green dye dilution method (CO DD ) in patients with COPD. Methods: Fifty paired CO measurements were concurrently obtained using the two methods from 10 patients (FEV 1 : 50.5 ± 17.5% predicted) at rest and during cycling at 25%, 50%, 75% and 100% peak work rate. Results: From rest to peak exercise CO IC and CO DD were strongly correlated (r = 0.986, P < 0.001). The mean absolute and percentage differences between CO IC and CO DD were 1.08 L/min (limits of agreement (LoA): 0.05-2.11 L/min) and 18 ± 2%, respectively, with IC yielding systematically higher values. Bland-Altman analysis indicated that during exercise only 7 of the 50 paired measurements differed by more than 20%. When data were expressed as changes from rest, correlations and agreement between the two methods remained strong over the entire exercise range (r = 0.974, P < 0.001, with no significant difference: 0.19 L/min; LoA: −0.76 to 1.15 L/min). Oxygen uptake (VO 2 ) and CO DD were linearly related: r = 0.893