Evaluation of cardiorespiratory coupling (CRC) usually requires the simultaneous recording of heart period (HP) variability, derived from the electrocardiogram (ECG), and respiration. ECG-derived respiration (ECGDR) exploits the cardiac axis movement due to respiration to estimate respiratory activity directly from the ECG. Since CRC indexes could theoretically be computed using ECGDR, a comparison with results obtained through a more precise monitoring of respiratory activity such as the respiratory flow (RF) is warranted. Therefore, a mixed unpredictability index (MUPI) of HP variability from respiratory dynamics, computed via local k-nearest-neighbor approach, was calculated using ECGDR and RF in patients with preserved functional capacity (PFC) and with reduced functional capacity (RFC) before and after cardiopulmonary exercise test (CPET) protocol. The MUPI computed from RF was found to be significantly increased in PFC patients after CPET protocol, while no effect could be observed when considering the ECGDR. Moreover, the correlation between the two MUPI indexes was limited. We conclude that indexes of CRC might require more direct measures of respiration than ECGDR to detect pathophysiological differences.