Overexpression of the Her-2/neu (HER2) oncogene is known to confer important prognostic and predictive value to patients with breast cancer. Controversy exists as to the best method for its determination caused primarily by the variable sensitivities of the different antibodies and interobserver differences, particularly in the group of breast cancers with borderline levels of expression of the protein product. This study was therefore designed to determine the status of the HER2 gene amplification in a group of breast carcinomas with low levels of overexpression. After an initial validation of our procedures, a series of 52 consecutive cases of formalin-fixed, paraffin-embedded breast cancers with low levels of overexpression and a series of 22 cases with no expression by immunohistochemistry were analyzed by fluorescence in situ hybridization (FISH), and the results correlated statistically. Amplification of the HER2 gene was observed in 16% of equivocal to weakly positive cases. Those that were amplified showed low levels of amplification with ratios less than 4.5 and a characteristic scattered pattern of distribution of HER2 signals in the FISH assay. In addition, heterogeneity was noted in two cases in the amplification of the HER2 gene within the same tumor samples with pockets of amplified tumor cells amidst nonamplified tumor cells. In cases without amplification, a statistically significant number showed chromosome 17 polysomy. In conclusion, equivocal to low levels of HER2 overexpression in breast cancers are associated, in the majority of cases, with chromosome 17 polysomy and a corresponding increase in the HER2 gene numbers. True gene amplification is present in only a minority of cases. FISH analysis should be used for confirmation of gene amplification. Prior screening and selection of appropriate immunohistochemistry-positive areas for FISH analysis may prove beneficial.