The accurate determination of HER-2 in invasive breast cancer has become a critical issue, particularly in the context of the results of recent trastuzumab (Herceptin((R))) adjuvant trials. This multicentre study evaluated inter-observer reproducibility in interpretation of HER-2 immunostains performed in different laboratories according to their in-house technique. A total of 74 HER-2 immunostains were evaluated by 16 pathologists and by a central review committee. As determined by central review, the HER-2 score was 0 in 33 cases (44%), 1+ in 10 cases (13%), 2+ in 9 cases (12%) and 3+ in 23 cases (31%). The overall kappa value was good (kappa=0.75). Agreement was excellent for the 0/1+ group (kappa=0.85) and for the 3+ group (kappa=0.82). As expected, the score 2+ group showed poor agreement (kappa=0.38). A quality assurance process showed that ring studies and adherence to national guidelines greatly improve inter-observer reproducibility.