Ventricular repolarization duration (VRD) is controlled by neural regulatory system same way as heart rate and, thus, also VRD varies in time. Traditionally, VRD variability is assessed by determining the time differences between successive R and T-waves, i.e. RT intervals. We have recently proposed a method based on principal component regression (PCR) for quantifying RT variability. The main benefit of the method is that it does not necessitate T-wave detection. In this paper, the noise sensitivity of the PCR based method is evaluated by examining the effect of simulated Gaussian noise on the spectral characteristics of the estimated RT variability series.