BACKGROUND/AIM There is increasing popularity for athletes to use heart rate variability (HRV) to tailor training. A time-efficient method is HRV assessment during deep sleep. The aim was to validate the selection of deep sleep segments identified by RR-intervals with simultaneous electroencephalography (EEG) recordings and to compare HRV parameters of these segments with those of standard morning supine measurements. METHODS In 11 world class alpine- skiers, RR-intervals were monitored during ten nights and simultaneous EEGs were recorded in 2-4 nights. Deep sleep was determined from the HRV signal and verified by delta power from the EEG recordings. Four further segments were chosen for HRV determination, namely a 4-h segment from midnight to 4 am, and three 5 min segments: one just before awakening, one after awakening in supine position and one in standing after orthostatic challenge. Training load was recorded every day. RESULTS A total of 80 night and 68 morning measurements of 9 athletes were analyzed. Good correspondence between the phases selected by RR-intervals versus those selected by EEG was found. Concerning root-mean-squared-difference of successive RR-intervals (RMSSD), a marker for parasympathetic activity, the best relationship with the morning supine measurement was found in deep sleep. CONCLUSIONS HRV is a simple tool for approximating deep sleep phases and HRV measurement during deep sleep could provide a time-efficient alternative to HRV in supine position.