To assess the genetic background of Liza affinis in the coastal waters of China, the sequence variation in 712 bp of the partial mitochondrial DNA COⅠgene of 84 individuals from 4 provinces was analyzed. In total, 18 polymorphic sites defined 21 haplotypes and low levels of genetic diversity (H_d = 0.82300.0400, pi=0.00230.0002) were detected. Individuals from different sites were intertwined together in the parsimony network, suggesting no geographical clustering. AMOVA analysis of various groupings (sampling sites, seas, division of the Taiwan Strait and Qiongzhou Strait) detected no significant genetic differentiation, as evidenced by pairwise fixation indexes (F_(st) = -0.1940 ~ 0.0141, P > 0.05), and the percentage of variation (-0.6%~0.46%) among groups. In the SAMOVA analysis, the Dongtou population was treated as one group, and the others were treated as another group; the percentage of variation among groups was 2.87%, congruent with the results of F_(st) and AMOVA analysis. The pattern of no lineage structure and no obvious genetic differentiation in L. affinis populations in coastal waters of China might have been caused by: 1) Passive transport of larvae and juveniles by marine currents, or active feeding migration and spawning aggregation of adults. 2) Recent population expansion. Significant negative Fus Fs for all populations (FS = -20.3900, P = 0), and obvious unimodal nucleotide mismatch distribution suggested that L. affinis in coastal waters of China had experienced population expansion and the deduced expansion time was approximately 134.199~14.911 kaBP; i.e., in the late Pleistocene, which might have been caused by the rise and fall of sea-levels with dramatic climatic changes during glacial-interglacial cycles. The Dongtou population should be given priority protection as it has the highest genetic diversity (H_d = 0.80800.1130, pi = 0.00210.0006) among the populations that were studied. Because the study only involved limited L. affinis populations in China, COⅠ might be a more conservative molecular marker than evolutionarily faster control region. Furthermore, maternally inherited mtDNA is independent of nuclear inheritance, and may not completely represent the evolutionary history of a species. Thus, future studies should incorporate more sampling sites across the species range using both nuclear and mtDNA markers together. A better understanding of the genetic background of L. affinis will provide a sound scientific basis for the preservation and sustainable utilization of its germplasm resources.