In an editorial 18 months ago, Wynford-Thomas raised a series of important issues concerning the application of antibodies for the immunohistological assessment of p53 protein.'Of particular note was the range of circumstances in which detectable p53 expression might not correlate with the presence of mutation of the p53 gene, whether giving rise to false-positive or falsenegative results. Since the publication of that important cautionary note, many papers have appeared on the topic of p53 immunohistochemistry and in this issue of the journal, another two papers are published which rovide some further insight into these points? 3 Moreover, since Wynford-Thomas's editorial, considerable progress has been made in our understanding of p53, and other caveats to the immunohistological assessment of p53 have also been recognized. The relationship between molecular abnormalities of p53 and neoplasia has been well reviewed elsewhere and there is strong evidence that abnormalities of p53 represent the most common molecular change in human Such abnormalities can be detected in a number of ways. For example, chromosomal changes may be detected directly by fluorescence in-situ hybridization or indirectly by conventional RFLP analyses. Mutations can be demonstrated by sequencing of the commonly mutated exons or inferred by SSCP analysis. Finally, abnormalities of the protein can be investigated by immunochemical methods including immunohistochemistry, a key point being the clear association between increased p53 protein stability and mutation. 236" To be of value in clinical practice, immunohistological assessment of p53 should provide clinical relevant information, particularly with regard to diagnosis or prognosis. The idea that there is, in general, a correlation between detectable expression of p53 protein and neoplasia comes from retrospective and descriptive studies of archival pathological materia18, 9 supported by some experi-