The ERBB2 transmembrane tyrosine kinase receptor is both a prognostic marker and a therapeutic target in breast cancer. Accurate determination of ERBB2 status is a prerequisite for the establishment of prognostic significance and for the selection of ERBB2-overexpressing breast tumours for specific treatment. Unfortunately, there is no complete agreement on how this determination should be performed. This study has compared four methods of assessment of ERBB2 status. Two global, extraction-based methods using real-time quantitative PCR on DNA (Q-PCR) or RNA (RQ-PCR) and two non-global, tissue-based methods, ie fluorescence in situ hybridization (FISH) and immunohistochemistry (IHC), were used. The 94 breast cancers tested were enriched in cases scoring 2+ using the IHC scoring system currently in use and for which the actual ERBB2 status remains ill defined. To determine the best parameters and reagents for assessment, two protocols for FISH and five anti-ERBB2 antibodies were used, and both FISH and IHC were carried out on the same tissue microarrays (TMAs). It is shown that the combination of the two tissue-based methods gives the best results. The use of either PCR-based method did not improve the results significantly. A new combined IHC score based on the association of two selected anti-ERBB2 antibodies (HercepTest trade mark and TAB250) and a dual scale for improved assessment of ERBB2 protein expression, particularly in 2+ cases, are proposed.