In an attempt to standardize the immunohistochemical detection of ERBB2, we evaluated six normal breast samples and 58 breast carcinomas for (1) the effect of four different fixatives (Bouin's fluid, buffered formalin, alcoholic formalin, and methacarn), and (2) the sensitivity and specificity of eight commercially available antibodies. ERBB2 gene copy numbers and RNA transcripts were measured by Southern and Northern blot analysis in all samples. The BT-474 and MCF7 cell lines were processed in the same way as the tissue samples and used a s references of activation or the normal state of the ERBB2 gene, respectively. Complete concordance between immunohistochemistry and molecular biology was observed in 43/58 cases (74 per cent). However, fixation and processing protocols significantly affected the reactivity of the antigenic determinants. The most consistent results were obtained with NCLCB11 on frozen sections and with Tab250 on paraffin-embedded sections. For prospective studies, the use of alcoholic formalin in association with a highly specific antibody (Tab250, 9G6, NCLCB11, or OA-11-854) gave the best results. Methacarn appears to be a reliable fixative, but was studied in only 28 cases. For retrospective studies, the most important parameter to take into account is the sensitivity of the antibody. Thus, with Bouin's fixation, 3B5 was the only reliable antibody, whereas with buffered formalin, Tab250 gave the most consistent results. This study will help inter-laboratory comparisons, and in determining the prognostic significance of ERBB2 expression in breast carcinomas using a standardized methodology.