Many trials using DNA microarrays have been reported for various human malignancies, but an efficient molecular diagnostic system has yet to be established. Here, we adopted a high throughput quantitative PCR-array system based on adaptor-tagged competitive PCR (ATAC-PCR), as a novel technique for gene expression profiling of hepatocellular carcinoma (HCC). This PCR-array contained 3,072 genes derived from three different cDNA libraries, including 298 additional known genes suspected to be involved in hepatocarcinogenesis. Using this PCR-array with 20 pairs of liver tissues (20 HCC, 20 surrounding nontumor liver), we identified a total of 117 genes differing in expression levels in the two liver tissues. Hierarchical clustering analysis and principal component analysis with these genes revealed distinct gene expression patterns in the HBV-positive group and the HCV-positive groups. Among 117 genes, only 7 (GPAA1, TMEM9, FACL4, ADFP, MAWBP, PACE4, FOS) were common to both groups. In conclusion, this PCR-array analysis with an appropriate set of genes is considered useful for gene expression profiling of HCC, and we identified some genes which may play a common key role in hepatocarcinogenesis.