Abstract: The Chinese mitten crab (Eriocheir sinensis) is one of the most widely cultured aquatic species in China and has high economic and practical value. However, its development has been severely constrained by germplasm degradation and inbreeding caused by long-term unregulated breeding practices. Therefore, genetic improvement through selective breeding has become an effective approach to address these challenges.To improve the efficiency of genomic selection and germplasm resource evaluation in the E. sinensis, this study aimed to develop a medium-density and high-accuracy whole-genome single nucleotide polymorphism (SNP) liquid-phase genotyping chip. Based on whole-genome resequencing data from 136 E. sinensis individuals, high-quality SNPs were identified through parameter filtering and quality control and subsequently used for probe design and synthesis. In the preliminary validation of the SNP array, three resequenced individuals used in array development and ten individuals randomly selected from 96 genotyped samples were analyzed, including two technical replicates. Genotypes generated by the SNP array were compared with corresponding whole-genome resequencing results to assess concordance. A production-scale evaluation was then conducted using 96 E. sinensis individuals with two technical replicates, and the resulting genotyping data were used to assess chromosomal distribution and genotyping performance of SNP loci. Liquid-based SNP array was used to genotype 724 E. sinensis individuals. Genomic estimated breeding values for color traits were predicted using the BayesB model, and the prediction performance was assessed in terms of accuracy and bias. ① In the preliminary validation experiment, the average call rate of 13 samples was 99.33%, with call rate differences between two technical replicates of 0.04% and 0.03%, respectively, and the average concordance rate between chip genotyping and resequencing results for three samples was 93.25%; ② In the production-scale test, SNP call rates ranged from 98.61% to 99.54%, and genotype concordance rates between two technical replicates were 99.78% and 98.76%, respectively; ③ SNP markers on the chip were evenly distributed across 72 chromosomes, with an average SNP density of 25.5 SNPs/Mb; SNP density across chromosomes ranged from 20 to 32 SNPs/Mb and showed a positive correlation with chromosome length; ④ Quality control indicators, including minor allele frequency (MAF) and Hardy–Weinberg equilibrium (HWE), demonstrated the high quality of SNP markers on the chip, with a total of 21 680 high-quality polymorphic loci identified;⑤The genomic selection results for color traits showed that the prediction accuracy of the four color parameters ranged from 0.1535 to 0.5576, while the prediction bias ranged from 0.6917 to 1.0025. Among these traits, redness (a^*) exhibited the highest prediction accuracy, indicating that the SNP array has good potential for practical application in breeding programs. The 30 K liquid-based SNP array developed in this study for the E. sinensis exhibits a high detection rate, high genotyping accuracy, and strong technical stability. It meets the requirements for genome-wide selection and germplasm resource evaluation in E. sinensis, and provides a reliable tool for molecular breeding, germplasm identification, and population genetic studies.