This paper compares two methods to estimate heart rate variability spectra, i.e., the spectrum of counts and the instantaneous heart rate spectrum. Contrary to Fourier techniques based on equidistant sampling of the interbeat intervals, the spectrum of counts and the instantaneous heart rate spectrum are based on non-equidistant sampling: the values are determined at R-wave moments. A consequence of the non-equidistant occurrence of the R-peaks in a heart rate signal is the appearance of the sidebands of the harmonic components of the mean heart rate in the spectra. These sidebands contaminate the signal components in the spectrum. The sideband distortion in the instantaneous heart rate spectrum was found to be smaller than in the spectrum of counts. Simulations using the IPFM-model were made to quantify this difference. On the basis of these simulations, sideband distortion appeared to be dependent on the mean heart rate, the modulation depth and the modulation frequency.