A series of 520 cases of non-Hodgkin's lymphoma (NHL; 374 of B-cell, 130 of T-cell, 5 of non-B/non-T-cell, and 11 of undetermined phenotype) was analyzed for the presence of Epstein-Barr virus (EBV) using RNA in situ hybridization (RISH). The aims of the study were to assess the frequency of EBV-encoded small nuclear RNAs 1 and 2 (EBER), abundant immediate early RNAs (BHLF), and latent membrane protein-1 (LMP-1) in cases covering the entire histologic spectrum of NHL, and to analyze whether EBV status had prognostic relevance with regard to patient survival. EBER positivity was found in 25 of 374 (7%) B-NHL and 40 of 130 (31%) T-NHL (P < .00005) cases, but in only 1 of 16 cases with non-B/non-T-cell or undetermined phenotype. Among T-NHL cases, EBER positivity was confined to angioimmunoblastic, lymphadenopathy-like lymphoma (11 of 13 cases, 85%), Lennert's lymphoma (five of seven cases, 71%), and pleomorphic T-NHL (24 of 67 cases, 36%). Mycosis fungoides, lymphoblastic, and CD30-positive anaplastic large T-cell NHL cases were consistently EBV-negative. Double-labeling by RISH and immunophenotyping demonstrated the presence of EBV in neoplastic T cells, but no CD21 expression was found in the EBER-positive T-NHL cases. LMP-1 was expressed in 12 of 40 (30%) EBER-positive T-NHL and 5 of 25 (20%) EBER-positive B-NHL cases. For both T- and B-NHL, no correlation was found for EBER positivity and age, sex, clinical stage, or serum level of lactate dehydrogenase (LDH) at diagnosis. However, in T-NHL but not B-NHL, EBER positivity correlated with the presence of constitutional symptoms and a poor performance score (PS < 1; scale, 0 to 4). EBER status did not have any prognostic significance in B-NHL, but it had a negative prognostic impact in high-grade T-NHL (7-year survival of EBER-negative v EBER-positive cases: 33% v 14%; P = .01). A multivariate analysis including all B- and T-NHL of intermediate-/high-grade histology showed that EBER positivity in T-NHL was one of the three most significant factors recognized by the final prognostic model, only surpassed by PS greater than 1 and age greater than 67 years, and more powerful than B symptoms, an elevated LDH, or disseminated disease (clinical stage greater than II). We conclude that patients with EBV-positive T-NHL have a very poor clinical outcome, that EBV status should be considered as additional useful information in the classification of T-NHL, and that EBV-positive T-NHL should be treated as a separate entity in the future.