Carcinogenicity of chemicals in our environment is one of the most important health hazards to humans. Recently, a microarray-based short-term prediction system for the hepatocarcinogenicity of chemicals, named CARCINOscreen(®), was developed. Although the system is a promising tool reported to have an ability to predict hepatocarcinogenicity in rats with 92.9% accuracy, it requires specialized equipment and skilled bioinformatics approaches for data analysis. Therefore, we attempted to develop a quantitative PCR (qPCR)-based system as an alternative to microarray-based CARCINOscreen(®). Finally, an optimized gene set consisting of four predictive genes (Abcb1b, Eprs, Map3K8, and Igh-6) was selected from among 3,150 combinations of candidate gene sets. The results of training- and validation-phase trials showed that the qPCR-based alternative to the microarray-based CARCINOscreen(®) could predict the hepatocarcinogenicity of chemicals in rats with 82.8%-86.4% accuracy. One of the predictive genes, Abcb1b, a member of the ATP-binding cassette protein superfamily and multi-drug resistance-associated protein, and the results of this study may indicate a close relation of this gene to the carcinogenicity of chemicals. The prediction performance of the qPCR-based CARCINOscreen(®), as well as its user-friendliness and cost efficiency, suggests that this method is promising for application to primary health hazard assessment. Thus, the qPCR-based CARCINOscreen(®) is considered as a promising tool for predicting the carcinogenicity of chemicals.