There is an increasing clinical demand for HER2 analysis in breast cancer, especially since the release of trastuzumab. The authors assessed the ability of immunohistochemistry to detect HER2 overexpression in invasive mammary carcinomas (IMC) using five antibodies. Paraffin-embedded samples of 86 IMCs (T2N0) were used to compare the immunohistochemical overexpression of HER2 using two polyclonal antibodies (HercepTest [DAKO] and A0485 [DAKO]) and three monoclonal antibodies (CB11 from two different laboratories, Biogenex and Novocastra, and 4D5 [Genentech]). All immunostainings were scored according to the FDA-approved HercepTest recommendations. The HercepTest-positive cases were compared with gene amplification by FISH (Oncor Inform, Ventana). The HercepTest was positive in 31 of the 86 cases (36.1%). The DAKO antibody A0485 was positive in 25 of the 66 (37.8%). Monoclonal antibody 4D5 was positive in only 15 of the 86 cases (17.4%). There was almost total agreement in results between the two CB11 antibodies: 25 of the 86 positive cases (29.1%). All cases positive for CB11 or 4D5 were HercepTest positive. Most of the HercepTest 2+ cases were negative when using either monoclonal antibody. FISH was positive in 19 of the 20 HercepTest 3+ cases and negative in 5 HercepTest 2+ cases. Three CB11-2+ cases showed no amplification by FISH. In three FISH-positive cases the immunohistochemistry showed no overexpression by all antibodies used. These findings suggest that immunohistochemistry may be used reliably as a primary methodology for evaluating HER2; however, the use of polyclonal antibodies may not be adequate to assess HER2 overexpression. CB11, regardless of the manufacturer (Biogenex or Novocastra), showed better concordance with FISH (kappa=0.83) than did the polyclonal antibodies.