At present, analysis of heart rate variability (HRV) is becoming widely used as a clinical or research tool. Supported reliability studies for HRV measurement are, however, still limited. The main purpose was to perform an assessment of the absolute and relative reliability of HRV parameters from short-term recordings by means of orthoclinostatic stimulation and to investigate, whether there is a difference in repeating the retest immediately or after several days. The study group consisted of 99 participants (mean age 22 +/- 1.24 years). Standard HRV indexes were computed: PT (total spectral power), PHF (high frequency spectral power), PLF (low frequency spectral power) and LF/HF. Absolute reliability was assessed by the standard error of measurement and 95% limits of agreement; relative reliability was assessed by the intraclass correlation coefficient. There was also an estimate of the sample size needed to detect the mean difference >/= 30% of the between-subject standard deviation. In conclusion, a large random variation (within individuals) of HRV parameters was revealed, regardless of whether the retest was repeated immediately or with an interruption. For most HRV parameters (particularly in the immediately repeated test-retest), however, random variation represents a limited portion of the between-subject variability.