This paper discusses the effect of atrioventricular conduction time (AVCT) on the short-term Heart Rate Variability (HRV) by computing HRV parameters using intervals between the onsets of successive P waves (PP time series) for three groups: normal, arrhythmia and sudden cardiac death (SCD) patients. A very precise wavelet transform based ECG delineator was developed to detect PP, PR and RR time series. Mean PR variation in arrhythmia and SCD group was found to be significantly high as compared to the normal group. It was observed that when PR variations in arrhythmia and SCD cases crossed a certain threshold, RR variability no longer provided a very accurate estimate of HRV. In such cases, PP variability was able to provide a better assessment of HRV.