The c-erbB-2 gene codes for a putative transmembrane protein, similar in structure to the epidermal growth factor (EGF) receptor. Amplification of the gene has been described in a variety of human adenocarcinomas and is particularly well documented in breast carcinoma. It has been suggested that amplification is indicative of poor prognosis and, as such, is comparable with lymph node status as a predictor of clinical outcome. This study examines the suggestion indirectly by an immunohistochemical technique. Archival tissue from 195 patients with primary breast carcinoma was stained with the polyclonal antibody 21N, raised to amino acids 1243-1255, the C-terminus of the predicted amino acid sequence of the c-erbB-2 protein. Up to 10 year verified follow-up data were available on all patients. Staining compatible with significant amplification was observed in 17 patients. Using the chi-squared test for trend a significant correlation was found between staining and grade (P = 0.04) but not with either node or receptor status. No significant association was found between staining and clinical outcome although there was a tendency for patients with stained tumours to have a worse prognosis. A Cox regression analysis was used to adjust for node status and grade and still no correlation was revealed between staining and prognosis. However a study of this size in which only a small number of patients have been found to have stained tumours does have wide confidence limits. Comparable staining observed in in situ and infiltrating components of tumours suggests that amplification is an early event in carcinogenesis. Similar staining in primary and subsequent metastatic lesions was also noted. It is considered that further studies at both the DNA/mRNA and protein levels are required to confirm the significance of c-erbB-2 amplification in human breast carcinoma.