Respiratory sinus arrhythmia (RSA) may improve the efficiency of pulmonary gas exchange by matching the pulmonary blood flow to lung volume during each respiratory cycle. If so, an increased demand for pulmonary gas exchange may enhance RSA magnitude. We therefore tested the hypothesis that CO2 directly affects RSA in conscious humans even when changes in tidal volume (V(T)) and breathing frequency (F(B)), which indirectly affect RSA, are prevented. In seven healthy subjects, we adjusted end-tidal PCO2 (PET(CO2)) to 30, 40, or 50 mmHg in random order at constant V(T) and F(B). The mean amplitude of the high-frequency component of R-R interval variation was used as a quantitative assessment of RSA magnitude. RSA magnitude increased progressively with PET(CO2) (P < 0.001). Mean R-R interval did not differ at PET(CO2) of 40 and 50 mmHg but was less at 30 mmHg (P < 0.05). Because V(T) and F(B) were constant, these results support our hypothesis that increased CO2 directly increases RSA magnitude, probably via a direct effect on medullary mechanisms generating RSA.