Objective: To determine the role of mitochondrial COⅠ gene on classification and identification of species, a total of 39 single individuals of 9 species pertaining to 2 genera of Cynoglossidae in the Yangtze Estuary were barcoded by COⅠ, sequenced and compared with that of other Cynoglossidae species recorded in GenBank (Table 1). Methods: Total genomic DNA was extracted from each scale sample using the classic phenol / chloroform extraction method. ~650 bp COⅠ fragments were amplified using the primers, i.e., F1: 5’-TCA ACC AAC CAC AAA GAC ATT GGC AC-3’, R1: 5’-TAG ACT TCT GGG TGG CCA AAG AAT CA-3’. Each PCR amplification was performed in a total volume of 50μl of PCR mixture. PCR products were purified and then sequenced in both forward and reverse directions using an ABI PRISMTM 3730 XL Automated Sequencer. DNA sequences were aligned with ClustalW using default parameters. Base composition, variable and parsimony informative sites were determined using MEGA 5.0. Neighbor-joining (NJ) and Maximum parsimony (MP) phylogenetic trees were constructed for COI haplotypes (Kimura 2 Parameter substitution model, K2P; 1000 bootstraps pseudoreplications) using MEGA 5.0. Results: Using the MEGA 5.0 software for statistical analysis, the averaged AT content is higher than the GC content (Table 2). The GC content of codon position 1 averaged 53.8% (51.8%-57.3%), that of position 2 for 42.0%, and that of position 3 ranged from 28.1% to 37.8% in average of 32.4% (Table 4). The transitionsal pairs (si) was slightly more than the transversional pairs (sv), with the ratio (R=si/sv) was 1.45 (Table 3). Analysis of the frequency of amino acids in COⅠ gene encoding protein, which showed that the highest frequency of amino acid is leucine, and the lowest frequency of amino acid is tryptophan (Table 5). The average K2P distances pairwise-species and within-species were 0.191 and 0.003, respectively (Table 6). The K2P distance pairwise-species was 63.7 times of that within-species. According to the MP and NJ trees for all 39 sequences, it demonstrated that Cynoglossinae in the Yangtze Estuary is a monophyletic group (Fig. 1). However, the phylogenetic relationships revealed by the COⅠ sequence analyses were not consistent with those inferred from morphological classification. In contrast with morphological classification, Paraplagusis japonica of Rhinoplagusia and Cynoglossus robustus of Cynoglossoides were placed in a same clade with support of a high bootstrap value (Fig. 1). The species in the sub-genera of Areliscus clustered on an independent branch. But, synonymic phenomena existed in the following two groups of species: C. abbreviatus and C. purpureomaculatus, C. lighti and C. joyner, with the pairwise-species of 0.002 and 0.007, respectively (Table 6). Conclusions: Our results highlight that the information from COⅠ sequences not only can filter out the synonym of the same species, but also be able to carry out effective identification for Cynoglossidae species, which further showed that mitochondrial COⅠ is feasible as the classification barcode.