We have previously shown that hypocapnia triggers Cheyne-Stokes respiration with central sleep apnea (CSR-CSA) in patients with congestive heart failure (CHF). Nasal continuous positive airway pressure (NCPAP) may attenuate CSR-CSA in patients with CHF and CSR-CSA. Accordingly, we hypothesized that attenuation of CSR-CSA by NCPAP would be related to an increase in PCO2. Therefore, we examined the effect of NCPAP on the frequency of apneas and hypopneas, transcutaneous PCO2 (PtcCO2), and minute volume of ventilation (VI) in 12 consecutive patients with CHF and CSR-CSA during stage 2 sleep. A control group of six patients, who did not receive NCPAP, was also studied. In the control group, there were no changes from baseline to 1 mo in the frequency of central apneas and hypopneas, mean PtcCO2, mean VI, or mean SaO2 during stage 2 sleep. In contrast, from baseline to 1 mo the NCPAP group experienced a decrease in the frequency of apneas and hypopneas (58.7 +/- 5.2 to 23.2 +/- 6.0/h of sleep, p < 0.001), an increase in mean PtcCO2 (34.6 +/- 1.4 to 40.8 +/- 1.1 mm Hg, p < 0.001), a reduction in mean VI (8.1 +/- 1.0 to 5.2 +/- 0.5 L/min, p < 0.01) and an increase in mean SaO2 (91.6 +/- 1.1 to 95.0 +/- 0.5%, p < 0.025) during stage 2 sleep while on 10.2 +/- 0.5 cm H2O nasal CPAP. We conclude that likely mechanisms through which NCPAP reduces CSR-CSA are by increasing SaO2 and raising PaCO2 during sleep toward or above the apneic threshold.