The transfer of cytotoxic effector cells reduces the risk of relapse after allogeneic BMT. Two murine leukemia cell lines, A20 (B lymphocytic) and WEHI-3 (myelomonocytic), were used to investigate antileukemic effector mechanisms operating independently from graft-versus-host disease (GVHD). Different results were obtained with the two leukemia models. After injection of A20 cells, the majority of Balb/c recipients treated with syngeneic BMT died due to leukemia relapse (89%). The transplantation of MHC-matched DBA marrow resulted in chronic GvHD but did not reduce the risk of relapse (86%). Grafting of MHC-mismatched (but GvH-nonreactive) marrow cells from (C57xBalb)F1 hybrids, however, led to a significantly lower relapse rate (47%, p < 0.05). In vitro testing revealed that F1 cells but not Balb/c or DBA cells exert NK cell activity against A20. The elimination of NK 1.1-positive cells from the graft reduced the antileukemic activity of (C57xBalb)F1 marrow against A20 in vivo. After injection of WEHI-3 leukemia cells, syngeneic BMT cured most of the recipients (62%) and transplantation of (C57xBalb)F1 marrow provided no additional benefit. In contrast to unmanipulated Balb/c and (C57xBalb)F1 cells, which showed no NK activity against WEHI-3 in vitro, IL-2 treated effector cells were highly cytotoxic. Transfer of IL-2 preincubated grafts significantly decreased the relapse rate of WEHI-3 (19 vs. 38%) after syngeneic and allogeneic BMT. Our data indicate that GvL activity can be separated from GvHD. In our murine model, GvL activity appears to depend more on the donors NK/LAK cell activity than on the presence or absence of GvHD.