Despite the apparent complexities of vocabulary and techniques involved in nucleic acid hybridization, these methods should become important new weapons in the pathologist's armoury. The particular strengths lie in the analysis of genes whose protein products defy detection either because they are absent (due, for example, to cellular controls as found in viral latency), to a point mutation, or to gene deletion (as is often the case with cellular oncogenes). However, the mere application of this technology will not solve all diagnostic problems. One must be aware, particularly in applying in situ hybridization to a new system, of the artefactual binding of the probes, and suitable control and duplicate experiments should be performed. In addition, it is vital to verify the identity of the probes and the specificity of the reaction by filter hybridization. The latter procedure may seem unwieldy, but is, in fact, no more complex than the recommended procedures for immunohistochemistry, as discussed by Matthews (this symposium). Sadly, for both hybridization and immunological detection, operator inexperience has often prevented such checks of specificity and as a result many spurious results populate the scientific literature, for example a recent controversy over the detection of Human T cell Lymphotrophic Virus Type 1 in multiple sclerosis patients. With careful application, however, these techniques will permit the detection of viral molecules under conditions where traditional electron microscopy/histology have failed, and may reveal possible viral aetiologies for a range of hitherto non-viral diseases.