A major goal in human health risk assessment is the identification and management of chemicals that may cause cancer in human populations. The current gold-standard for assessing chemical carcinogenicity is the two-year rodent cancer bioassay, which is animal, time, and resource intensive. The use of toxicogenomics for chemical risk assessment was first proposed over 15 years ago because of its potential to produce toxicologically relevant data more quickly, using fewer animals, and at a lower cost than the two-year cancer bioassay. While the two-year cancer bioassay produces a detailed inventory of chemical-dependent lesions, toxicogenomics analyzes chemical-dependent changes to global gene expression. Moreover, toxicogenomics provides comprehensive mechanistic data that are not obtained using standard tests. In this thesis quantitative, predictive, and mechanistic approaches were applied to a toxicogenomic case study of the rodent hepatocarcinogen furan. Female B3C6F1 mice were exposed for three weeks to non-carcinogenic or carcinogenic doses of furan. The dose response of a variety of transcriptional endpoints produced benchmark doses (BMDs) similar to the furan-dependent cancer BMDs. Bioinformatic analysis of disease datasets showed strong similarity between global gene expression changes induced by furan and those associated with the appropriate hepatic pathologies. The molecular pathways that were enriched in the liver following furan exposure facilitated the development of a molecular mode of action (MoA) for furan-induced liver cancer. Finally, transcriptional changes in formalin-fixed and paraffin embedded (FFPE) samples were compared to high quality frozen samples in order to evaluate whether archival samples are a viable option for toxicogenomic studies. The advantage of using FFPE tissues is that they are very well characterized (phenotypically); the disadvantage is that formalin degrades biomacromolecules, including RNA. FFPE samples as old as three decades were shown to be feasible for toxicogenomics studies using a ribo-depletion RNA-seq protocol. Taken together, this case study demonstrates the utility of toxicogenomics data in human health risk assessment and the potential of archival FFPE tissue samples, and identifies viable strategies toward the reduction of animal usage in chemical testing.