The majority of single nucleotide polymorphisms (SNPs) found in the coding region (cSNPs) are single base substitutions that may or may not lead to amino acid substitutions, most of which are related to diseases. Some cSNPs may prove useful for their potential links to functional cSNPs via linkage disequilibrium mapping. We have selected 48 cSNPs located in the coding regions of 25 genes to construct the cSNP chip. These genes are harbored in the high frequency loss regions of the chromosome 1p and 8p and related with apoptosis, cell cycles, signal transduction, oncogene, tumor suppressor genes and so on. All of the cSNPs can lead to amino acid substitutions except TP73 (rs1801174). The PCR products amplified from 31 hepatocellular carcinoma (HCC) specimens were labeled with Dig-dUTP and then hybridized with the cSNP chips. The results showed that there was no hybridization signal when there was more than one site of mutation in the amplification sequence, indicating that the cSNP chip had a high sensitivity. The statistic data of the SNP (MT, homozygous and HT, heterozygous) in the HCC patients with different phenotypes (HBV+/−, differentiation stage, family history positive or negative, tumor size) indicated that the number of MT was distinctly different between patients with positive HBV and negative HBV. The MT and HT numbers of all the 48 cSNPs were significantly different between low differentiation and high differentiation HCC patients. The numbers of MT and HT were not different between positived and negative family history groups and between tumor size >3 cm and ⩽ 3 cm groups. The study results provided useful information for understanding the molecular mechanisms of HCC development.