Graft-versus-leukemia (GVL) is a major component of the overall beneficial effects of allogeneic bone marrow transplantation (BMT) in the treatment of leukemia. Although several clinical trials have suggested a direct relationship between GVL effects and acute and chronic graft-versus-host disease (GVHD), it is not yet known whether GVL can be separated from GVHD. However, several investigations in murine models of human leukemia indicate that the two may be at least partially separable. Moreover, analysis of clinical data from the International Bone Marrow Transplant Registry suggest that allogeneic BMT may be more advantageous than syngeneic BMT, regardless of the GVHD. Likewise, T lymphocyte depletion is associated with an increased incidence of relapse, independently of GVHD. Recent investigations in murine leukemia suggest that GVL-like effects may be inducible following syngeneic BMT by recombinant cytokines with no overt GVHD. Taken together, current data in experimental animals and man suggest that GVL may be at least partially separable from GVHD. Hence, further understanding of effector and target cells of GVL as well as our ability to induce antitumor effector cells, especially those that are MHC nonrestricted, may lead to new approaches for potentiating anti-tumor effector mechanisms without inducing severe, clinically overt GVHD. Successful attempts in these directions may also lead to improved results following autologous BMT as a result of activation of GVL-like effects by recombinant cytokines that are capable of activating effector cells with anti-leukemic activity in vivo, such as recombinant human IL2, alpha interferon or perhaps a synergistic combination of factors.