Her-2/neu (H2N) status in breast carcinoma has been considered a prognostic factor that may have therapeutic implications; however, the correlation between H2N overexpression and gene amplification has not been completely defined. A consecutive series of ductal carcinomas (34 invasive and 7 in situ) were analyzed by fluorescent in situ hybridization for H2N gene and chromosome 17 copy number using touch preps of intact cells and by immunohistochemistry, using three different commercial antibodies to H2N protein (Zymed, clone 31G7; Ventana, clone CB11; and Dako, polyclonal) in corresponding formalin-fixed, paraffin-embedded tissue sections. Gene amplification was classified as unequivocal if more than five signals were present in more than 80% of the counted nuclei and absent if more than 80% of the nuclei counted contained two or fewer gene copies. Cases that did not fulfill the above criteria were considered equivocal for amplification. Immunostaining was classified as follows: 0 = no staining; 1+ = faint, incomplete membranous pattern; 2+ = moderate, complete membranous pattern; 3+ = strong membranous pattern. Of the 34 invasive tumors, 10 (29%) had unequivocal gene amplification. Furthermore, all had more than 10 copies of the gene in more than 60% of the counted nuclei. An additional nine cases (26%) had equivocal amplification, which was usually the result of chromosome 17 aneuploidy (seven of nine) or heterogeneity. With the Zymed and Dako antibodies, all tumors with 3+ staining had unequivocal gene amplification and all cases with 2+, 1+, or 0 staining were negative or equivocal for gene amplification. With the Ventana antibody, all cases with 3+ staining had unequivocal gene amplification, but two cases with unequivocal amplification by fluorescent in situ hybridization exhibited 1+ staining. Moderate (2+) H2N staining was observed in one case, three cases, and five cases with the Ventana, Dako, and Zymed reagents, respectively, and did not correlate with H2N gene copy number. Discordance between H2N and chromosome 17 copy number was not a useful means of defining amplification. Two cases of ductal carcinoma in situ with the Zymed antibody and two with the Dako antibody showed 3+ staining despite lack of unequivocal gene amplification. We conclude that (1) strong H2N immunostaining is highly associated with gene amplification, although there is minor variation in sensitivity between different antibodies; (2) a subset of breast carcinomas (3 to 15%) demonstrate moderate H2N staining without evidence of amplification, and it is unclear whether they represent highly sensitive staining or are a subset of cases that show overexpression without amplification; (3) gene amplification, as detected by fluorescent in situ hybridization, is associated with at least 10 gene copies per nucleus, and lower gene copy duplication (3 to 4/nucleus) is frequent, usually the result of chromosome 17 polysomy, and not associated with high-level overexpression; (5) overexpression of H2N without amplification may be more frequent in ductal carcinoma in situ, implying a different role in the biology of preinvasive versus invasive neoplasm.