The HER2 gene is amplified and overexpressed in 25-30% of breast carcinomas. Assessment of HER2 status for prognosis and treatment of breast cancer patients can be performed by immunohistochemistry and/or fluorescence in situ hybridization (FISH). To develop a testing algorithm for HER2 in breast cancers, we used FISH analysis to determine the HER2 gene copy number and immunostaining to detect the p185 protein. Interlaboratory, interobserver, and intermethod variabilities of immunohistochemistry were assessed. In 24 invasive breast carcinomas, the indices of HER2 status obtained by FISH and by a reference laboratory's DAKO HercepTest (immunostain) gave an overall concordance of 96%. The reference laboratory's stained slides were re-interpreted by an in-house panel of pathologists; the interpretation differed in one case. The panel's interpretations were concordant with the FISH results in all 24 cases. Interobserver variability for the panel's immunohistochemistry interpretations was assessed using three different immunostaining methods on 70 slides. The numerical (0-1+, 2+, 3+) scores showed greater variability among observers than did the overall positive/negative results. One pathologist reported inconsistent results in >30% of the slides evaluated. Borderline scoring of 1-2+ was reported in 18 slides (23%) by at least one observer. Incongruent interobserver immunohistochemistry scores, leading to different positive and negative interpretations, were obtained with 5 slides (7%). The majority of consensus positive cases exhibited strong membrane staining (3+). The majority of consensus negative cases scored as 0. Based on these observations, we developed a testing algorithm that maximizes the benefits of FISH and immunohistochemistry, providing physicians with accurate results for appropriate clinical care.