Abstract: Objective To evaluate the accuracy of a novel PCR-independent mtDNA deep sequencing method developed in previous studies compared with the traditional mtDNA sequencing based on long fragment PCR ( IrPCR ) method. Methods Ten peripheral blood samples were simultaneously subjected to parallel experiments using the both methods: novel and traditional lrPCR-based NGS ( next gene sequence) method. The sequencing results obtained from the two approaches were compared, and the accuracy was evalua- ted by combining the information from 2 population databases ( MitoMAP and mtDB). Results A total of 380 variants were detected in the 10 samples. Of these，351 were detected by the both methods, and 29 were detected exclusively by single method. The 8 variants in the 29 inconsistent variants which appeared in 2 or more samples were further analyzed. The results showed that 2 variants which were only repeatedly detected by the new method were recorded in the population database. However, none of the 6 variants detected only by the traditional method were recorded in the population databases. Conclusion Compared with the traditional IrPCR -based NGS meth- od, the newly developed method has exhibited a similar detection capability for both of homoplasmic and high-level heteroplasmic varia- ntsThe new method exhibited higher accuracy for low heterogeneity variation detection and has shown advantages of higher accuracy in detection for low level of heteroplasmic variants.