Oestrogen receptor (ER) status is the only biochemical predictive factor which is routinely measured in breast carcinomas. ER gene mutations can profoundly change the biochemical activity of the protein. If these occurred in vivo, they could be expected to affect breast cancer risk or phenotype, such as endocrine responsiveness. However, no mutations of significance have been described in breast carcinomas. In contrast, numerous variant forms of ER have been reported at the mRNA level. Most of these appear to be due to aberrant exon splicing which results in predicted protein products whose activities range from dominant positive to dominant negative. In some instances, these mRNA variants have also been demonstrated in normal tissue (breast and others). Their biological and clinical significance might be profound, but remain to be established because of a lack of evidence for their existence at the protein level. On the currently available data, routine analysis for ER mutants and variants is not justified.