Objective To analyze a large number of Cumulative Paternity Index (CPI) data in Single-parentage identification with stochastic simulation method. Methods Using the software in Identifiler, PP18D, AGCU EX20, PP21, AGCU 21+1 system, 1 million groups of STR genotyping of parent-child were analyzed. The average rate of very low CPI (RVLCPI), the median of PI in every loci (MPI), and the average PI of 39 STR loci (PIsolo) were counted out. And calculation formula of the number of STR loci that need more testing (N) was derived. Results RVLCPI in Identiifler, PP18D, AGCU EX20, PP21, AGCU 21+1 system were 24.9%, 6.5%, 1.4%, 0.7%, 0.000 1%. On the premise of inclusion relations between every kits, the number of autosome STR loci is inversely proportional to RVLCPI. PIsolo was 1.843 2. There was a signiifcant positive correlation between MPI and PEduo (r=0.956, P<0.001). The formula of the number of STR loci that need more testing was N=?log1.843 2(10 000/P)?. Conclusion The result and the formula in this paper are effcient in single-parentage identiifcation.