It is more than 80 years since the first attempts were made to immunize human cancer patients against their own tumors (for review see reference 1). As yet, there are no immunological manipulations of proven value in terms of ensuring long-term cure, though there are randomized control studies demonstrating short-term prolongation of survival in patients with acute myelogenous leukemia receiving bacille Calmette-Guerin (BCG) 1 with or without aUogeneic leukemic blasts (for review see reference 2). In experimental animals, particularly inbred mice, it has been considerably easier to demonstrate tumor-specific antigens, both by serological and cell-mediated techniques and to prolong survival by immunizing against these antigens (3). Despite a welter of publications, there are few studies in human tumor systems where the concept of tumorspecific antigens has been clearly established and universally confirmed. In particular, though many authors have demonstrated lymphocyte-mediated cytotoxicity against tumor lines in both normal individuals and patients with tumors (4, 5), evidence for Tlymphocyte-mediated autologous tumor cytotoxicity has been very difficult to establish.In vitro assays of transplantation rejection using the mixed lymphocyte culture and cell-mediated lympholysis tests, have provided clear insight into the nature of antigenic determinants necessary for generation of cytotoxic T cells. The pioneering studies of Eijsvoogel et al. (6) on HLA recombinant families demonstrated the need for two types of antigens, the first type coded for by genes closely associated with those of the HLA D locus cause lymphocyte activation, whereas the second class of determinants, closely associated with the serologically defined HLA A, B, and C locus antigens, acts as the target for the cytotoxic T lymphocytes. Eijsvoogel et al. (7) also showed that production of cytotoxic T lymphocytes was possible even if the lymphocyte-activating determinant and serologically defined antigens were present on different cells. Zarling et al. (8) applied this principal to study leukemia patients and then reported one patient where stimulation of remission lymphocytes by mitomycin C inactivated autologous myeloblasts and a third party inactivated allogeneic lymphocytes in short-term culture led to the production of cytotoxic cells against autologous blasts but not the remission lymphocytes.We report and confirm this observation in a series of 14 patients with acute myelogenous leukemia and investigate the nature of the effector cell, the helper * Present address: