The reactivity of eight preferential B-cell (L26, 4KB5, and KiB3) and T-cell (polyclonal CD3, Leu22, MT-1, UCHL-1, and OPD4) antibodies which detect paraffin-resistant antigens was examined by a threestep immunoperoxidase technique in 111 formalin-fixed, paraffin-embedded diffuse aggressive non-Hodgkin's lymphomas (NHLs) to determine the optimal panel for accurate lineage assignment. L26 (CD20) and polyclonal CD3 (CD3) were the most sensitive (> 95%) and specific (100%) antibodies. They identified the B-or T-cell lineage correctly in 106 (95%) cases. The five L26-negative, polyclonal CD3-negative cases included all three precursor B lymphoblastic NHLs and two (one B and one T) diffuse large-cell NHLs. Immunostaining with the second most sensitive preferential B-cell (4KB5) and T-cell (Leu22) antibodies correctly identified the lineage in two addiThe recent advent of a variety of monoclonal and polyclonal antibodies reactive with paraffin-resistant B-and T-cell-associated antigens has markedly increased our ability to analyze lymphoid proliferations in routinely processed tissues. However, the immunophenotypic analysis of non-Hodgkin's lymphomas (NHLs) with these antibodies in paraffin-embedded tissue sections has often been considered inferior to the immunophenotypic analysis of viable cells in suspension and frozen tissue sections. In some cases, however, the use of routinely processed, paraffin-embedded tissue is preferable to frozen tissue-for instance, when accurate morphologic identification of the malignant cells is necessary or when it is important to preserve unfixed tissue for other studies. Furthermore, because specimens often arrive in the pathology laboratory already in fixative, fresh tissue may not be available for immunophenotyping. In addition to the histologic subtype and stage, the B-or T-cell lineage of an NHL often is clinically important because tional NHLs, but one false-positive result occurred. Preferential B-cell antibody KiB3 reacted with two precursor B lymphoblastic NHLs. Use of additional paraffin-reactive antibodies did not increase the number of NHLs assigned to the correct cell lineage. In conclusion, it appears that a two-tiered approach, with a first-line panel consisting of L26 and polyclonal CD3, followed by 4KB5 and Leu22 in nonlymphoblastic NHLs and by KiB3 in lymphoblastic NHLs, represents the most efficient method of correctly identifying the B-or T-cell lineage of diffuse aggressive NHLs by paraffin tissue section immunohistochemistry.