We have studied the intratumor HER-2/neu heterogeneity in 78 consecutive and population-based primary invasive breast carcinomas. Within the invasive component, heterogeneity was detected in only 1 of 78 tumors. In 48 tumors (62%), we found both in situ and invasive components in analyzed tissue sections. Twelve of these 48 tumors had a difference of at least 2 arbitrary units in the in situ compared with the invasive part of the tumor with regard to the HER-2/neu status analyzed by HercepTest (immunohistochemistry). Eight of these 12 tumors were reanalyzed with fluorescent in situ hybridization and immunohistochemistry with and without a new Automated Cellular Imaging System. In this limited material, immunohistochemistry in combination with the Automated Cellular Imaging System seemed to have a better correlation with fluorescent in situ hybridization than immunostaining analyzed manually. In conclusion, HER-2/neu expression is not seldom heterogeneous in invasive compared with in situ components within a tumor. This finding should be considered in the choice of evaluation method. To avoid heterogeneity as a confounding factor in HER-2/neu analyses, detection methods such as immunohistochemistry and fluorescent in situ hybridization, which can provide evaluation in a preserved tissue architecture, should be used. Perhaps the intratumor HER-2/neu heterogeneity can explain some of the unexpected failures of trastuzumab therapy.