Transfer of T-cells to heavily irradiated, H-2-compatible mice frequently leads to a high incidence of lethal graft-versus-host disease (GVHD). Lymphoid cells depleted of Thy1+ cells fail to cause GVHD. Studies with a variety of different, H-2-compatible, strain-combinations suggest that minor, histocompatibility antigens (minor HA) are the main targets for eliciting lethal GVHD. Experiments in which T-cells are negatively selected to minor HA by acute blood-to-lymph recirculation through irradiated hosts have indicated that the T-cells eliciting GVHD to minor HA, are H-2-restricted. In H-2-compatible hosts, the donor T-cells recognize the minor HA of the host and become temporarily trapped in the lymphoid tissues for 1-2 days; during this stage of negative selection, the donor T-cells entering the lymph are specifically devoid of cells able to elicit GVHD against the host, minor HA on further transfer. When the selection host is H-2-different with respect to the donor T-cells, by contrast, the T-cells ignore the host, minor HA and negative selection fails to occur. The T-cells recirculate normally and are unimpaired in their capacity to elicit GVHD on further transfer. By the use of various H-2-recombinant mice as selection hosts it has been shown that, as for T-cells exerting cell-mediated lympholysis (CML) to minor HA in vitro, the T-cells which elicit lethal GVHD to minor HA comprise two distinct subsets of H-2-restricted cells. One subset recognizes minor HA in the context of H-2K (or K end) molecules whereas the other is specific for minor HA-plus-H-2D. Curiously, in marked contrast to the findings on CML responses in vitro, no evidence has been found that H-2I-restricted T-cells contribute to GVHD, either as effector cells or as helper cells. Purified populations of Lyt 1-2+ T-cells have potent GVHD activity, whereas Lyt 1+2- cells fail to cause GVHD. Studies with various types of bone-marrow chimeras suggest that in the induction phase, T-cells recognize minor HA only on lymphohematopoietic cells. In the effector phase, by contrast, non-marrow-derived cells appear to be the main targets of attack. Although the pathogenesis of GVHD is poorly understood, the lethal form of the disease probably reflects the penetration of mucosal surfaces by pathogenic organisms, perhaps as the result of direct destruction of epithelial cells by minor HA-specific cytotoxic lymphocytes. Direct support for this notion has yet to be obtained.