OBJECTIVE: To investigate whether epileptic seizures could be predicted or detected by means of spectral analysis of heart rate variability (HRV). METHODS: Six patients with temporal lobe epilepsy (4 females, 2 males) participated in the prospective pilot study while enrolled for video/EEG monitoring (24 h/day, 2-4 days). ECG was continuously recorded and 30 min seizure-sessions (25-30 min pre-seizure to 30 sec-5 min post-seizure onset) and 30 min non-seizure-sessions (day- and night sessions for each patient, as control) were chosen for further HRV-analysis. Low frequency (LF) (0.04-0.15 Hz), High frequencies (HF) (0.15-0.40 Hz), LF/HF, LF/(LF+HF) and reciprocal HF-power was determined using continuous FFT- spectral analysis of 64 R-R interval windowing with maximum overlapping. RESULTS: Six seizures were recorded and analyzed from three patients (2 females, 1 male). All of the analyzed EEG-correlated seizures showed reciprocal HF-power peaks between 10 sec pre seizure-onset and 24 sec post seizure-onset with peak amplitudes 2.96-93.63 times higher than control maximum peak. For the other parameters we could not find significant difference between seizure and non-seizure sessions. CONCLUSION: Specifically high reciprocal HF-power peaks suggest suppressed parasympathetic activity just around seizure-onset time. Seizure detection using HRV-analysis seems to be a promising method for non-invasive seizure detection in the early phase of the clinical event (even preceding the onset).