The Shannon entropy Hk in symbolic dynamics analysis of time series has less data demand and can be complemented easily, and therefore is applied to heart rate variability (HRV) analysis. However, the criterion has not been established as to the parameter α, which is used during the transformation into symbols. Like all other physiological signals, great differences between individuals as well as nonstationarity are usual in HRV. Therefore, to achieve a robust and consistent analysis, the parameter α should be considered together with the mean and standard error of the series. In this paper, the integrative effects on Hk of the three parameters were studied firstly in simulation time series, and it was suggested that under certain conditions, the clusters of Hk approach a convergent upper boundary named Hk-up, which reflects the intrinsic dynamics of the time series. Then, in the heart beat interval series analysis of 15 subjects, the abilities of Hk-up to alleviate the impacts brought by both difference between individuals and nonstationarity were testified.