Periodic breathing with central apneas during sleep is typically triggered by hypocapnia resulting from hyperventilation. We therefore hypothesized that hypocapnia would be an important determinant of Cheyne-Stokes respiration with central sleep apnea (CSR-CSA) in patients with congestive heart failure (CHF). To test this hypothesis, 24 male patients with CHF underwent overnight polysomnography during which transcutaneous PCO2 (PtcCO2) was measured. Lung to ear circulation time (LECT), derived from an ear oximeter as an estimate of circulatory delay, and CSR-CSA cycle length were determined. Patients were divided into a CSR-CSA group (n = 12, mean +/- SEM of 49.2 +/- 6.3 central apneas and hypopneas per h sleep) and a control group without CSR-CSA (n = 12, 4.9 +/- 0.8 central apneas and hypopneas per h sleep). There were no significant differences in left ventricular ejection fraction, awake PaO2, mean nocturnal SaO2, or LECT between the two groups. In contrast, the awake PaCO2 and mean sleep PtcCO2 were significantly lower in the CSR-CSA group than in the control group (33.0 +/- 1.2 versus 37.5 +/- 1.0 mm Hg, p < 0.01, and 33.2 +/- 1.2 versus 42.5 +/- 1.2 mm Hg, p < 0.0001, respectively). Neither group had significant awake or sleep-related hypoxemia. In addition, CSR-CSA cycle length correlated with LECT (r = 0.939, p < 0.001). We conclude that (1) hypocapnia is an important determinant of CSR-CSA in CHF and (2) circulatory delay plays an important role in determining CSR-CSA cycle length.