In the post genome era it will soon be possible to associate a specific tumor type with a specific gene expression profile and to define each molecular lesion characteristic of any given cancer. It is intuitive that a successful therapeutic strategy for cancer should aim at blocking the aberrant biochemical activity triggered by the oncogene or the lack of tumor suppressor gene activity that ultimately leads to full-blown neoplastic transformation. However, an attractive alternative approach entails the blockade of the transcriptional consequences of such oncogenic activities irrespective of their original biochemical nature, thus antagonizing the key transcriptional events underlying cancer pathogenesis in any specific neoplastic cellular population. This approach is now rendered possible by major advances along several lines of investigation: (i) the possibility of analysing gene expression through high throughput methods; (ii) a more detailed knowledge of the regulatory regions and of the transcription factors that control gene expression also facilitated in the future by a comprehensive whole genome comparative analysis of these regulatory sequences; (iii) the ability of modulating gene expression at the single gene level through various approaches both pharmacological and biochemical; (iv) the opportunity of directly antagonizing the aberrant activities of oncogenic transcription factors through a detailed knowledge of their abnormal transcriptional function; (v) the possibility of validating, in vivo, in animal models the relevance for neoplastic transformation of specific transcriptional events as well as of testing the efficacy of 'transcription therapy' in faithful animal models of human cancer. Here, we will review the facts, the existing applications and the hypothesis underlying such therapeutic modality for cancer therapy.