The clinical interest in HER-2/neu is related to trastuzumab, a drug used to treat patients with invasive breast carcinoma overexpressing the HER-2/neu protein. It is very important to correctly identify those patients who may benefit from trastuzumab by accurate assessment of the HER-2/neu status. Of the various methods available, the Dako Herceptest for immunohistochemical assay is considered the most reliable to reach this goal. The aim of this study was to investigate within a group of Italian laboratories the reproducibility of the results of HER-2/neu assessment by means of the Dako scoring system on slides stained with the Herceptest kit. This study was also conceived as the continuation of one of our previous studies, which was similar in its aims but different in the classification criteria adopted. Our results show that, whereas the intra-observer reproducibility was generally satisfactory, the interobserver reproducibility was not. Moreover, our findings confirm that the two extreme classes (0 and 3+) are more easy to identify than the other two and that the Herceptest does not allow to discriminate optimally between scoring classes 2+ and 3+. These findings are relevant in clinical practice where the treatment choice is based on categories defined by this assay, suggesting the need of adopting educational programs and/or new reference materials to improve the assay performance.