Summary This SERVE‐HF (Treatment of Predominant Central Sleep Apnea by Adaptive Servo Ventilation in Patients With Heart Failure) sub study analysis evaluated polysomnography (PSG) data in patients with heart failure with reduced ejection fraction (HFrEF) and predominant central sleep apnea (CSA) randomised to guideline‐based medical therapy, with or without adaptive servo ventilation (ASV). Patients underwent full overnight PSG at baseline and at 12 months. All PSG recordings were analysed by a core laboratory. Only data for patients with baseline and 3‐ or 12‐month values were included. The sub study included 312 patients; the number with available PSG data differed for each variable (94–103 in the control group, 77–99 in the ASV group). After 12 months, baseline‐adjusted respiratory measures were significantly better in the ASV group versus control. Although some between‐group differences in sleep measures were seen at 12 months (e.g., better sleep efficiency in the ASV group), these were unlikely to be clinically significant. The number of periodic leg movements during sleep (PLMS) increased in the ASV group ( p = 0.039). At 12 months, the respiratory arousal index was significantly lower in the ASV versus control group ( p < 0.001), whilst the PLMS‐related arousal index was significantly higher in the ASV group ( p = 0.04 versus control). ASV attenuated the respiratory variables characterising sleep apnea in patients with HFrEF and predominant CSA in SERVE‐HF. Sleep quality improvements during ASV therapy were small and unlikely to be clinically significant. The increase in PLMS and PLMS‐related arousals during ASV warrants further investigation, particularly relating to their potential association with increased cardiovascular risk.