Two methodological aspects of heart rate variability (HRV) are tested. Heart rate variability and RR interval variability (RRV) are compared in 109 healthy subjects using 24-h Holter recordings. Autoregressive power spectrum analysis was performed, and the power in three frequency bands was evaluated by two different methods: the peak power and the mean power within the band. The power in the low frequency (LF) and high frequency (HF) ranges showed greater changes between day and night values when RRV was analysed compared with HRV. The mid frequency (MF) range power showed no significant change at night with RRV analysis compared with a significant decrease with HRV analysis. The different pattern of changes between day and night values when using RRV or HRV analysis is explained by the non-linear nature of the relationship between RR interval and heart rate and the lower heart rate at night. Therefore, depending on the heart rate, there may be amplification or attenuation of fluctuations when transforming from one heart rate variability method to another. When comparing peak power and mean power measurements within the frequency bands, the changes in LF and MF powers were similar. However, HF mean power with RRV analysis showed less change at night than the peak power and no change with HRV analysis. The explanation lies in the fact that respiratory fluctuations at night are much more regular, resulting in a sharp high peak in the power spectrum. The peak power is therefore higher at night, whereas the total power changes much less or remains unchanged.(ABSTRACT TRUNCATED AT 250 WORDS)