OBJECTIVE: The aim of this study was to investigate the variations and the circadian rhythm of multi-domain parameters of heart rate variability (HRV) in patients with drug-resistant epilepsy. METHODS: Time domain HRV parameters of 24-hour data, frequency and non-linear domain HRV parameters were calculated using a 24-hour ambulatory electrocardiogram (ECG) from subjects in a resting and awake state (5-min data from 9:00-10:00 a.m.). Circadian fluctuation (24 one-hour segments) and the night (00:00 - 05:00)/ day (07:30 - 09:30) ratio of each HRV parameter were measured in (i) refractory epilepsy (n = 51) and (ii) healthy, age and sex matched controls (n = 50). Cosinor fit method was used to quantify the pattern of the circadian rhythm of HRV. RESULTS: Compared to healthy controls, patients with drug-resistant epilepsy had significantly decreased time domain (mean_RR, STDRR, RMSSD, pNN50, HRV_triangular_index), frequency domain (VLF_power, LF_power, HF_power, Total_power) and non-linear domain (Poincare SD(1), SD(2), D(2)) HRV measurements. There were also overall reductions in amplitude and the baseline of circadian curves of HRV parameter in the patients group, especially nocturnal inhibition of cardiac autonomic modulations was stronger than that in daytime. The majority of differences in HRV values between the two groups reached a maximum in early morning (usually 5:00 or 6:00 a.m.). No significant differences were found in the night/day ratios of multi-domain HRV measurements between patients and healthy controls (p = 0.862). CONCLUSIONS: The significant suppression of HRV parameters in patients with refractory epilepsy may reflect the inhibitions of both sympathetic and parasympathetic activities. This is notably the first time that marked circadian diversity of HRV measurments between the patient and control groups were found and quantified by cosinor fit method. The difference in amplitude increased during night with peak occurrence between 5:00 or 6:00 a.m. These findings may be linked to cardiac autonomic dysfunction and subsequent sudden unexpected death in epilepsy (SUDEP), which is reported to often occur during night or early morning. Taken together, HRV parameters may be potential bio-markers for predicting SUDEP.