Diagnosis of primary gastrointestinal T-cell lymphomas is often problematic because lymphoma may not be suspected clinically or on resection, and tissue may not be frozen for immunophenotyping. Furthermore, ulceration and inflammation and the polymorphous character of the lesions makes evaluation difficult. Only approximately 150 cases have been reported, fewer than 20 from North America. We have tried to establish a reliable approach to the diagnosis of gastrointestinal lymphomas of T-cell phenotype in routinely processed tissue. Sections from five primary gastrointestinal T-cell lymphomas were stained with a panel of 13 antibodies reactive in routinely fixed, paraffin-embedded tissue. These included antibodies to pan-T- and pan-B-cell antigens (CD3, CD20), B- and T-cell-associated antigens (CD43, CD45R0; CDw75, CD74), antigens expressed by activated T-cells (HLA-DR, CD30), leukocyte antigens (CD45, CD15), and macrophage markers (MAC-387, HAM-56). All stained positively with T-cell markers MT-1 and Leu-22, four with UCHL-1, and three with anti-CD3 polyclonal antibody. B-cell markers identified by L-26 and LN-1 were negative in all five, whereas LN-2 was expressed in two. Two expressed HLA-DR; all were Ber-H2 negative. Two had an abnormal phenotype: one was Leu-M1 positive, and one LCA negative. Ten B-cell gastrointestinal lymphoma controls were negative for MT-1, Leu-22, and CD3, and nine were negative for UCHL-1. Nine were positive for the B-cell marker L-26, eight for LN-2, and seven for LN-1. All tumors were negative for monocyte-macrophage markers. This antibody panel provides a reliable means for identifying gastrointestinal T-cell lymphomas in paraffin sections. Use of a panel is advisable because of variation in expression and preservation of antigens, and to detect abnormal phenotypes. Application of this approach may facilitate the diagnosis of gastrointestinal T-cell lymphomas both prospectively and in archival material, and thereby encourage studies of the behavior and treatment of these neoplasms.