Abstract: The kuruma shrimp (Marsupenaeus japonicus) includes two morphologically similar varieties, which are distributed mostly allopatrically but coexist in the northern South China Sea. In order to further analyze the phylogenetic relationship between the two phenotypic varieties of M. japonicus, the gene structure, genetic distance and codon preference of their mitochondrial whole genomes. The results showed that there were differences in gene overlap region, gene spacing and codon preference between the two genomes. The similarity of the nucleotide sequences of the mitochondrial protein-encoding genes was between 91.10% and 95%, and the similarity of amino acid sequences was between 96.15% and 100%. The sequence homology of A+T rich regions was only 82.60% and the divergence was 15.40%, indicating a high degree of genetic differentiation. The nucleotide sequence divergence of protein-coding genes between the two varieties was higher than that between Fenneropenaeus and Orconectes, while the divergence of amino acid sequence was smaller than that between the latter. The study found that the interspecific genetic distance based on cytb gene was more than 10 times the intraspecific genetic distance. The interspecific genetic distance based on cox1 gene was 14.6 times and 5.2 times of the intraspecific genetic distance for variety Ⅰ and variety Ⅱ, respectively. The Ka/Ks values of nd1, nd4 and nd5 genes were greater than 1, indicating positive selection. Phylogenetic trees were constructed based on the protein coding sequence of 20 Penaeus species belonging to 9 genera, and the result showed that each genus could be effectively differentiated. In the phylogenetic tree, the two varieties of M. japonicus were clustered first, and then clustered with Melicertus latisulcatus. In view of these differences between the two Marsupenaeus species, we believe that it is essential and urgent to establish a genetic database for each and reevaluate their suitable ecological conditions in order to improve species-specific culturing techniques.