Continuous positive airway pressure (CPAP) improves idiopathic central sleep apnea (ICSA), but the mechanisms are not completely understood. It has been demonstrated that some ICSA patients have an increased CO2 drive, expressed by the hypercapnic ventilatory response (HCVR). The aim of our study was to evaluate whether CPAP can decrease the HCVR in successfully treated ICSA patients. Also the central apnea mean duration (CMD) was studied. Nine patients were evaluated. Their apnea-hypopnea index at baseline was high, but decreased from 50 +/- 5 to 8 +/- 3, during 5 +/- 1 cm H2O CPAP. After one month CPAP high values were still found for HCVR (2.69 +/- 0.23 L.min-1.mmHg-1 at baseline, 2.71 +/- 0.33 L.min-1.mmHg-1 after CPAP). Sleep quality was only improved by a decrease in the arousal index (from 29 +/- 12 to 6 +/- 1 a night) (p = 0.04). Daytime arterial O2 partial pressure (PaO2) increased from 73 +/- 4 mmHg to 85 +/- 3 mmHg (p = 0.02), whereas daytime arterial CO2 partial pressure (PaCO2) remained unchanged (from 41 +/- 1 mmHg to 40 +/- 1 mmHg). CMD was 21 +/- 6 s before and 13 +/- 1 s after CPAP (p = 0.05). The presence of hypersomnolence decreased from in 89% to in 33% of the patients. It is concluded that CPAP treatment can induce a subjective and objective improvement in ICSA patients, with improved sleep-related respiratory indices, daytime PaO2, but without a change in CO2 drive. High chemical drives seem to be the cause and not the consequence of sleep disordered breathing and persist after treatment.