Seriola has three species that are highly similar morphologically and are difficult to distinguish. Therefore, it is necessary to distinguish them by molecular biology methods. In this study, we explored the applicability of cytochrome oxidase subunit I (CO I), cytochrome oxidase subunit II (CO II), and 16S rRNA in identification and evolutionary analysis of the three species of Seriola, namely, S. lalandi (Chinese, Australian and Japanese populations), S. dumerili, and S. quinqueradiata. Three pairs of primers were designed for PCR amplification and sequencing of CO I, CO II, and 16S rRNA sequences of the three Seriola species. Dnasp5.10 and Mega7.0 software were used for DNA sequencing, genetic difference analysis, and phylogenetic tree analysis. The results showed that the CO I, CO II, and 16S rRNA sequences had a significant A+T bias. The mitochondrial 16S rRNA sequence was the most conservative, with a variation rate of 5.06%. The average number of nucleotide differences (k) and nucleotide diversity (Pi) according to the CO I sequence were higher than those of CO II and 16S rRNA. CO II had the highest haplotype diversity index among the three genes, indicating its high genetic diversity. The CO I sequence had higher differentiation and more abundant genetic resources than the 16S rRNA and CO II sequences. Thus, CO I and CO II sequences could be used to identify the three Seriola species in China and effectively identify the populations of S. lalandi from the three different waters (Northeast, Northwest and South Pacific waters). In the Seriola species, the interspecific genetic distances of CO I and CO II were more than 10 times the intraspecific genetic distances. The phylogenetic tree analysis showed that each species had independent branches. S. lalandi and S. quinqueradiata had a closer relationship than the others in the tree. The phylogenetic tree constructed using the three gene sequences could reflect the evolutionary relationship among the species. These results indicated that CO I and CO II gene sequences could be used as effective DNA barcodes for species identification and population division of Seriola species and provide efficient molecular tools for classification identification, resource screening, and diversity protection.