Résumé : This study investigated the degree of interlaboratory agreement when HER-2/neu was evaluated by immunohistochemistry (IHC) on archival primary breast cancer samples. IHC for HER-2/neu was performed on the same archival tissue sections from 394 invasive primary breast cancers in two different laboratories. Both laboratories used the primary antibody NCL-CB11; however, different methods of immunostaining (antigen retrieval procedure and manual processing or no antigen retrieval and autostainer processing) as well as different scoring systems were used. Fluorescence in situ hybridization (FISH), considered as the correlation method for HER-2/neu status determination, was performed using the PathVysion kit and compared to the IHC results. Forty-eight of 394 analyzed tumors (12.2%) were scored as HER-2/neu positive in one laboratory, and 109 (27.7%) in the other laboratory where antigen retrieval was performed. Complete concordance in categorization of HER-2/neu status between the two laboratories was achieved in 333 of 394 cases (84.5%). FISH performed in 248 formalin-fixed samples revealed HER-2/neu gene amplification in 55/248 (22.2%). Concordance of FISH and IHC was found in 211/248 cases (85.1%) and 220/248 cases (88.7%) when the CB11 antibody was used without and with antigen retrieval, respectively. Both IHC methods generated similar rates of false results, but with different positive predictive values. Our data demonstrate that HER-2/neu evaluation by IHC is not a reproducible technique if there is no standardization of the procedure.