Purpose Adaptive servo-ventilation (ASV) is a therapy designed for patients with central sleep apnea (CSA) and Cheyne Stokes respiration. The aim of this study was to find predictors of ASV usage in patients with CSA in a routine sleep clinic cohort. Methods In this retrospective study, consecutive patients in whom ASV therapy was initiated at the University Hospital Regensburg between 2011 and 2015, were analyzed. Analysis included polysomnographies of diagnostic and ASV initiation nights, a phone questionnaire on ASV usage, readout of the ASV device 1 month after initiation ("early ASV usage," 1 month after ASV initiation), and the readout of the last month before a reappointment date set in 2015 ("late ASV usage," median 17 months after ASV initiation). Results In 69 consecutive patients, the mean early and late ASV usage per night was 4.8 +/- 2.5 h and 4.1 +/- 3.0 h, respectively. Seventeen months after initiation, 57% of patients used the device >= 4 h per night, and of those 91% reported a subjective benefit from ASV therapy. Early ASV usage was significantly associated with late ASV usage (univariable regression: Beta 0.8, 95%CI [0.6; 1.0]p< 0.001). In multivariable regression analysis, short duration of slow wave sleep (N3) during diagnostic polysomnography (Beta - 6.2, 95%CI [- 11.0; - 1.5];p= 0.011) and subjective benefit from ASV (Beta 174.0, 95%CI [68.6; 279.5];p= 0.002) were significantly associated with longer late ASV usage. Conclusion Early ASV usage predicts late ASV usage. In addition, low slow wave sleep before ASV initiation and subjective benefit from ASV may contribute to higher late ASV usage.