Molecular technology plays an ever-increasing role in academic pathology and frequent claims are made that these technical developments will revolutionize diagnostic histopathology. In reviewing the role and potential utility of molecular genetics in the clinical practice of tumour pathology, it is evident that claims have outweighed facts, at least to date, and while molecular data have helped to clarify specific small areas of tumour classification and have facilitated the diagnosis of certain relatively rare tumours (especially on limited tissue), the overall impact on cancer management has been small. In particular, molecular markers of prognosis or tumour response have yet to impact substantially on the role of conventional light microscopic examination. We believe that these shortcomings have more to do with the generally different goals of basic medical science and clinical practice. In order to optimize the benefit to patients of modern molecular pathology, these techniques need to be rigorously appraised in the clinical setting and properly analysed in terms of cost-benefit. This requires close collaboration between the basic research community and histopathologists. Suggestions that 'H&E' pathology will be superceded in the near future are both premature and unrealistic.