We compared the c-erbB-2 protein overexpression status detected by the HercepTestTM (DAKO A/S, Grostrup, Denmark) with another conventional immunohistochemistry system using an anti-c-erbB-2 rabbit polyclonal antibody (Nichirei Co., Tokyo, Japan) and with the c-erbB-2 gene amplification status detected by Southern blot hybridization in 101 surgically resected breast carcinomas. According to the criteria for overexpression, recommended by the manufacturer, c-erbB-2 overexpression by HercepTestTM was detected in 24 cancers (24%), comprising six score-2 tumors and 18 score-3 tumors. The level of agreement in judgment of the HercepTestTM, among three independent observers, was excellent (kappa = 0.845). C-erbB-2 overexpression by Nichirei's antibody and c-erbB-2 gene amplification were detected in 21% and 16% of cases, respectively, and their concordance with HercepTestTM scores of 2-3 was 89% and 90%, respectively. In the score-3 cases by Hercep TestTM only, the concordance rates with overexpression by Nichirei's immunohistochemistry and with gene amplification were slightly higher, 94% and 93%, respectively. Score-2 cases by HercepTestTM were mostly judged as negative overexpression by Nichirei's antibody and as no amplification by Southern blot hybridization. The present results showed that HercepTestTM score-3 detected c-erbB-2 overexpression almost optimally as well as the conventional methods and the score-3 breast carcinomas had clinical and biological implications. Further examination would be necessary to decide the significance of breast cancers of HercepTestTM score 2.