Because apnea length during periodic breathing varies according to the preceding increase in ventilation and reduction in PaCO 2, differences in the cycle length of periodic breathing among patients with normal and impaired cardiac function might be explained by the influence of lung-to-carotid body circulatory delay, as reflected by lung-to-ear circulation time (LECT), on hyperpnea length rather than on apnea length. It was therefore hypothesized that circulatory delay is an important determinant of periodic-breathing hyperpnea length but not apnea length. To test this hypothesis, LECT, periodic-breathing cycle length, apnea length, and hyperpnea length were compared in 10 patients with idiopathic central sleep apnea (ICSA), whose cardiac function was normal, as opposed to 10 with Cheyne-Stokes respiration and central sleep apnea (CSR-CSA) in association with congestive heart failure (CHF). As compared with ICSA patients, cycle length was significantly longer in patients with CSR-CSA (37.3 +/- 3.0 s versus 59.0 +/- 4.9 s, p < 0.005). This difference was due to significantly longer hyperpnea length in the CSR-CSA patients (16.7 +/- 2.8 s versus 36.7 +/- 3.4 s, p < 0.001), since apnea length was similar in the two groups. In addition, LECT was longer in the CSR-CSA patients (24.3 +/- 2.0 s versus 10.3 +/- 1.0 s, p < 0.001), and correlated strongly with cycle length (r = 0.88, p < 0.001) and hyperpnea length (r = 0.90, p < 0.001) but not with apnea length. LECT correlated inversely with cardiac output (r = -0.72, p < 0.006), indicating that LECT is a valid measure of circulatory delay. Thus, circulatory delay is an important determinant of hyperpnea length but not of apnea length in patients with ICSA and CSR-CSA.