Heart rate variability (HRV) spectrum can be calculated from the R-R or the sequence of instantaneous heart rates (IHRs). Because these signals are reciprocal, their use in the determination of the correlation of HRV with average HR may yield opposite results. AIM: The aim of this study is to reveal how HRV parameters correlate with HR depending on the kind of signal used and whether the normalization procedure or use of corrected signals affect the correlations. METHODS: Electrocardiographic recordings (512 beats) of 55 patients were considered in the analysis. RR and IHRs were calculated. Both signals were divided by their average values yielding the corrected RR and IHR. Heart rate variability spectra were estimated from these 4 kinds of signals. Total power (TP), low frequency (LF) and high frequency (HF) components, LF/HF ratio, and normalized values (ie, nLF, nHF) were calculated. RESULTS: Total power and LF estimated from RR correlated negatively with HR, but those calculated from IHR revealed a trend toward a positive correlation (respective correlation coefficients significantly differed, P < .01). The sign of correlation between HF and HR was the same regardless of the type of signal applied. The application of corrected signals made the results almost identical despite different signal origins (RR or IHR). The correlations of LF/HF, nLF, and nHF with HR were similar in all cases. CONCLUSIONS: The type of signal determines the sign of correlation among TP, LF, and HR. The parameters obtained from corrected signals, normalized quantities, and LF/HF reveal a consistent relationship with HR.