Amplification of Her-2/neu in breast carcinoma is associated with poor prognosis, short disease-free interval, and short survival time in both node-negative and -positive patients. Little is known about the starting point of amplification of Her-2/neu and how it progresses from benign to malignant breast lesions. We attempted to address these questions by evaluating amplification of Her-2/neu in benign, premalignant, and malignant lesions using fluorescence in situ hybridization (FISH). Twenty-six patients with Her-2/neu-overexpressing invasive ductal carcinomas (as judged by strong immunoreactivity with Her-2/neu antibody) and coexisting lesions of ductal hyperplasia (DH), atypical ductal hyperplasia (ADH), ductal carcinoma in situ (DCIS) in the vicinity of the invasive tumor (as judged by review of the hematoxylin and eosin-stained sections), as well as metastatic carcinoma in axillary lymph nodes (mets) were selected for this study. In the primary carcinomas, a close relationship was present between overexpression as detected by immunohistochemistry (IHC) and amplification as demonstrated by FISH (85% concordance). Among these patients, amplification of Her-2/neu in ADH was demonstrated in 7 of 13 cases with ADH, and in DCIS, in 21 of 22 cases with DCIS. There was no amplification in DH or normal ductal epithelium. Significantly, in all 12 patients with synchronous positive axillary lymph nodes, there was concordant amplification of Her-2/neu in the primary and metastatic carcinoma. Amplification was consistent in multifocal metastases, despite morphological heterogeneity in some patients. Amplification ratios increased from ADH to DCIS to invasive carcinoma (P <.01, ADH versus DCIS; P <.05, DCIS versus invasive cancer), but there was no difference in amplification ratios between primary cancers and synchronous axillary metastases (P >.05). We also evaluated Her-2/neu amplification in 21 patients without Her-2/neu overexpression in their primary carcinomas (as judged by absent immunoreactivity with Her-2/neu antibody). Three showed amplification in both primary and metastatic lesions, with a low amplification ratio (approximately 2). One patient had amplification in the primary tumor but not in an axillary metastasis. Two patients exhibited slight amplification in the metastatic carcinoma (ratios 1.6 and 2), but not in their primary cancers. This FISH study indicates that amplification of Her-2/neu can emerge de novo in any stage of the disease process, from ADH to metastatic lesions, but most often appears first in ADH or DCIS. The degree of Her-2/neu amplification increases with progression to invasive carcinoma, there being no further increase in synchronous metastasis. Our data suggest that amplification of Her-2/neu appears to be mainly involved in initiation of breast oncogenesis and that its role in progression of breast cancers is uncertain.