Hepatocellular carcinoma, the second leading cause of cancer death in China, is responsible for over 130,000 deaths every year. Until now, there have been very few screening tests for early diagnosis of hepatoma, which often delays early treatment. We have developed a modified representational difference analysis (RDA) technology and identified more than 200 upregulated genes in hepatoma, among which over 60% were later confirmed as truly upregulated genes in additional hepatoma samples. Approximately 20%–30% of these upregulated genes had detectable expression differences between normal and hepatoma samples in a mini-cDNA array based on the approximately 200-gene set. We built a cDNA array that represents a set of 11,000 human genes, including approximately 5,000 previously reported and 6,000 new EST clusters. By applying this cDNA array we further identified more than 800 upregulated genes in clinical hepatoma samples that were not detected by the previous RDA method. Combining the two methods significantly facilitated the discovery of hepatoma-related genes, as the cDNA array provides a broader view of gene expression level and subtractive cloning offers greater sensitivity in detection of low-abundance gene transcripts.