The human herpes virus Epstein-Barr (EBV) is clearly associated with African Burkitt's lymphoma and the undifferentiated from of nasopharyngeal carcinoma, although its role in oncogenesis is still poorly defined. Recently EBV has been implicated in other types of lymphomas, as well as in some nonlymphomatous neoplastic processes. Its possible association with human breast cancer has been investigated here. DNA from 91 cases of breast carcinoma and blood samples from the same patients were amplified with the PCR over a region in the EBV BamHIW major repeat sequence following a single-step amplification protocol. Nineteen samples (21%) were found to be positive; 10 samples of blood (only 3 of them from patients with EBV-positive tumors) were found by the adopted protocol to contain EBV DNA. Another series of PCR amplifications using primers covering a unique (nonreiterated) fragment in BamHIC encoding the EBERs (two short nonpolyadenylated RNAs generally highly expressed in cells latently infected with EBV) confirmed these findings. A good correlation between the two sets of experiments was observed, and only five differences in results were obtained on samples tested. In situ hybridization was carried out using BamHIW biotinylated DNA probes or EBER-1 digoxigenin-labeled riboprobes with the aim of confirming as well as localizing the signal to the epithelial cell. Twelve sections (63%) among the PCR-positive samples were found positive by in situ hybridization with the DNA probe, and six (31.5%) sections were found with the RNA probe. Twenty-one samples from benign breast tumors or normal breast tissue were used as controls for PCR amplification in this study, none of which was found positive. This is the first known report showing positive results for EBV in breast cancer. No statistical association was found in these studies between the presence of EBV and the histological type of the tumor, however. Its role therefore remains for the moment unknown, as well as does the significance of the association of EBV with only a subset of the cases.