Graft rejection has hampered the use of T cell depletion (TCD) in allogeneic bone marrow transplantation. A model of host-versus-graft (HVGR) and graft-versus-host reaction (GVHR) as two inversely related processes has been proposed. We investigated graft rejection rates in graft-versus-host-reactive and graft-versus-host-nonreactive situations in a rat and a mouse model. Model 1: LEW rats were pretreated with a fixed myeloablative dose of busulfan and increasing doses of the immunosuppressive cyclophosphamide. The animals received different doses of semiallogeneic GvH-nonreactive BM cells. Graft rejection rates were dependent on the bone marrow cell number transplanted and on the pretransplant immunosuppression. Graft rejection rates following transplantation of GvH-reactive CAP marrow and genetically GvH-nonreactive (CAP x LEW)F1 marrow were the same. In conclusion, there was no advantage with respect to engraftment for the GvH-reactive marrow. Model 2: In irradiated Balb/c mice, graft rejection rates following T cell-depleted and unmanipulated transplantation of GvH-reactive or GvH-nonreactive bone marrow grafts were identical. All experiments were done with graded numbers of BM cells and revealed a strong impact of the BM cell dose on engraftment. In our experiments the cell loss during the ex-vivo manipulation was approximately 50% and, in contrast to the clinical situation, we readjusted to the intended number after TCD. Our experiments demonstrate that neither GvHR nor T cells but the BM cell dose has a strong impact on engraftment of allogeneic bone marrow.