Genomics, proteomics and metabonomics are applied to toxicology either as stand-alone technologies or in combination, with the intention of providing a more efficient assessment of the potential side effects of new chemical entities. Two different approaches are taken: a predictive/proactive strategy based on a statistical analogy of 'signatures' of drugs to many known toxicant gene or metabolite fingerprints; and a mechanistic/reactive strategy based on the in-depth biological analysis of the gene, protein or metabolite profiles induced by one or a few compounds of interest. This article focuses on the advantages and disadvantages of these technologies, as well as the many hurdles associated with both these approaches in toxicology that have to be considered before applying them to the assessment of future drugs.