DNA microarray technology was developed as a tool for simultaneously measuring a number of gene expression changes, and has been applied for investigations of toxicity assessments of chemicals. In this study, we used a typical hepatotoxicant, thioacetamide (TA), to find correlations between the extent of hepatotoxicity and certain gene expression patterns or specific gene expression profiles. TA was intraperitoneally administered at high (400 mg/kg), medium (150 mg/kg) or low (50 mg/kg) dose (four rats per group) and then the serum and liver were collected at the indicated time (6, 12, 24, 36 and 48 h). Serum biochemical markers were measured and hepatic mRNA expression profiles were analyzed by a DNA microarray. Serum aspartate aminotransferase (AST) and alanine aminotransferase (ALT) were increased by TA-administration in a dose-dependent manner and reached the maximum at 24h. Hierarchical clustering analysis of all dosage groups revealed in 2 major clusters, distinguished by an early (6 and 12h) and a late (24, 36 and 48 h) phase. The early and late phase clusters were sorted in time- and dose-dependent manners. The major gene expression profile obtained by quality-threshold (QT) clustering analysis showed the same maximal toxic time as that estimated by the serum biochemical markers. The individual expression profiles of the candidate genes selected in our previous studies and the simultaneous gene expression patterns measured by five typical hepatotoxicants including TA also reflected the hepatotoxicity of TA. These findings suggest that the potential toxic effects appearing as gene expression changes are independent of the dosage of TA. This study suggested that the major gene expression profile estimated by QT clustering would be a sensitive marker of hepatotoxicity.