Clinical and experimental data suggest a role for the immune response in preventing leukemic relapses following allogeneic bone marrow transplantation the graft-versus-leukemia (GVL) effect. In the context of an allogeneic BMT, a number of different immune mechanisms mediated by donor cells may be responsible for the GVL effect. We have approached this question by using limiting dilution cultures of alloactivated human lymphocytes to analyze the in vitro allogeneic cytolytic response against fresh allogeneic leukemia. Initial results in the limiting dilution assays with split culture analyses demonstrated frequent alloreactive cytolytic T lymphocyte precursors that destroyed remission peripheral blood lymphocytes and leukemic cells from the allogeneic leukemic patient. These assays also demonstrated frequent lymphokine-activated killer (LAK) cell precursors that lysed both the LAK sensitive Daudi line and the allogeneic leukemia. In these experiments, isolated cultures also showed cytolytic activity directed against the allogeneic leukemic blasts without activity against remission PBL, or the LAK-sensitive Daudi cell line. Two T cell lines (ABL1 and ABL2) isolated from an LDA, demonstrated this form of specificity, mediating destruction specifically against the allogeneic acute lymphoblastic leukemic cells. Both cell lines ABL1 and ABL2 were CD3+, TCR alpha beta +, and CD4+. These 2 cell lines mediated little or no cytotoxicity against a large panel of other targets tested (natural killer sensitive and resistant cell lines, allogeneic PBL, and allogeneic fresh leukemic blasts). Antibody-blocking experiments revealed a role for the CD3-TCR receptor of both cell lines in lysis of leukemic cells; the CD4 and MHC class II molecules were clearly involved in the lysis by the ABL1 cell line. Specificity of recognition for the allogeneic leukemic blasts was further confirmed by unlabeled target competitive inhibition studies. The mechanism of the preferential lysis of leukemia by the alloactivated T cell lines described in this paper remains uncertain. Nevertheless, these leukemic-specific populations provide a means by which the human GVL effect may be further studied in vitro.