Previous studies have suggested that common breast cancers are associated with EBV. We used a highly sensitive quantitative real-time PCR method to screen whole tumor sections of breast cancers for the presence of the EBV genome. EBV DNA was detected in 19 of 92 (21%) tumors, but viral load was very low in positive samples (mean = 1.1 copy EBV/1000 cells, maximum = 7.1 copies EBV/1000 cells). Importantly, quantitative real-time PCR failed to detect the EBV genome in microdissected tumor cells from any case. Using a monoclonal antibody (2B4-1) reactive against the EBV nuclear antigen-1, we noted strong staining of tumor nuclei in a proportion of those breast cancers that had tested negative for the presence of the EBV genome. Because nuclear staining with the 2B4-1 antibody was previously observed more frequently in poor prognosis breast cancers, we examined a larger series of breast cancers with complete clinical follow-up. Strong punctate staining of tumor cell nuclei was observed in 47 of 153 (31%) breast cancers; 2B4-1-positive tumors were significantly more likely to be ER-negative (P < 0.0001), to be of higher grade (P = 0.001) and larger (P = 0.03), to involve more regional lymph nodes (P = 0.01), and to have higher Nottingham Prognostic Index scores (P = 0.0003). Conclusions are: (a) EBV can be regularly detected in whole sections of breast cancers but viral copy number is very low; (b) in these cases, tumor cells do not harbor virus; and (c) reactivity with the monoclonal antibody 2B4-1 is detectable in the absence of the EBV genome and is strongly associated with ER-negative breast tumors and with prognostically unfavorable disease. Additional studies should be directed to the identification of this protein and to elucidation of its role in breast cancer.