The aim of this study was to examine the dependence of heart rate variability (HRV) measures on RR interval length and to find out relationships between linear and nonlinear measures. The spectral powers in very low (VLF), low (LF) and high (HF) frequency regions, short-term scaling exponent alpha(1) and sample entropy (SampEn) were determined. All spectral powers increased with increasing RR interval length until they reached a plateau. Neighbouring spectral powers were strongly correlated. The largest fraction of the spectrum consisted of VLF (from about 40 to 95%) and the smallest of HF, although HF was most sensitive to changes in RR interval length. SampEn is also increased with increasing RR interval, reaching a plateau. The dependence of alpha(1) on RR showed a deflection point at 0.5 s. Nonlinear measures can be expressed by spectral powers: alpha(1) by a linear function of ln(LF/HF) and SampEn by a quadratic function of ln HF. We concluded that during the day an increase of HR up to 120 beats min(-1) was reflected in a reduction in HF and LF and to a smaller extent in VLF and by decreased complexity and increased correlation in RR interval series. In sleep, HRV measures are at a plateau. We suggest that below intrinsic value, HR is regulated mainly by changes of parasympathetic activity, reflected in linear and nonlinear HRV measures.