Good sleep is an indispensable factor for maintaining healthy life and sleep impairment is increased risk of morbidity and mortality among patients with chronic disorders. The standard method for diagnosing sleep disorder is polysomnography, but the development of a simpler method is desired due to its time and economic burden. In this study, we developed the assessment of sleep quality by nighttime electrocardiography (ECG) and examined its usefulness for risk stratification among chronic hemodialysis patients with end-stage renal disease. We introduced new ECG indices for estimating non-REM sleep period and the frequency of sleep disordered breathing in addition to the conventional risk markers of heart rate variability (HRV). We observed that mortality risk during a follow-up for 100 months was predicted by the markers of shortened total non-REM period and blunted cardiac response to sleep apnea independently of age, comorbidity including diabetes, and conventional HRV indices.