Adaptive servo-ventilation (ASV) is a novel method of ventilatory support designed for Cheyne-Stokes respiration (CSR) in heart failure. The aim of our study was to compare the effect of one night of ASV on sleep and breathing with the effect of other treatments. Fourteen subjects with stable cardiac failure and receiving optimal medical treatment were tested untreated and on four treatment nights in random order: nasal oxygen (2 L/min), continuous positive airway pressure (CPAP) (mean 9.25 cm H(2)O), bilevel (mean 13.5/5.2 cm H(2)O), or ASV largely at the default settings (mean pressure 7 to 9 cm H(2)O) during polysomnography. Thermistor apnea + hypopnea index (AHI) declined from 44.5 +/- 3.4/h (SEM) untreated to 28.2 +/- 3.4/h oxygen and 26.8 +/- 4.6/h CPAP (both p < 0.001 versus control), 14.8 +/- 2.3/h bilevel, and 6.3 +/- 0.9/h ASV (p < 0.001 versus bilevel). Effort band AHI behaved similarly. Arousal index decreased from 65.1 +/- 3.9/h untreated to 29.8 +/- 2.8/h oxygen and 29.9 +/- 3.2/h CPAP, to 16.0 +/- 1.3/h bilevel and 14.7 +/- 1.8/h ASV (p < 0.01 versus all except bilevel). There were large increases in slow-wave and rapid eye movement (REM) sleep with ASV but not with oxygen or CPAP. All subjects preferred ASV to CPAP. One night ASV suppresses central sleep apnea and/or CSR (CSA/CSR) in heart failure and improves sleep quality better than CPAP or 2 L/min oxygen.