The removal of T lymphocytes from intestinal allografts prior to transplantation would prevent graft-versus-host disease and might also weaken the unusually severe rejection response mounted by graft recipients. Ex vivo perfusion by monoclonal antibody-toxin conjugates represents a potentially ideal approach to achieve this goal. Monoclonal antibody-toxin conjugates were prepared by coupling the mouse anti rat CD5 antibody MRC OX19 to the A chain of ricin. The resultant conjugate contained 1 or 2 ricin A chain molecules per molecule of immunoglobulin and had high selective toxicity for rat T lymphocytes. Following ex vivo perfusion of the small intestine, the mesenteric lymph nodes and Peyer's patches were removed and the level of penetration of the MRC OX19 antibody was assayed by immunohistological techniques. In lymph nodes, there was ready access of the antibody to the medulla, and to a lesser extent to the B cell areas of the cortex. However, only the T cells in the peripheral regions of the paracortex were stained, suggesting that the paracortex was resistant to penetration by blood-borne antibody, and was being stained only by diffusion from the lymph node medulla. Some penetration of the antibody also occurred in the immediate vicinity of the postcapillary venules. In the Peyer's patches, staining was seen in the germinal centers, but not at all in the T cell areas. The addition of agents such as histamine to the perfusate to increase vascular permeability did not alter this picture. These studies suggest the presence of a potentially interesting resistance to penetration of the paracortex of the lymph node by blood-borne substances. Nevertheless, sufficient penetration occurred to influence favorably the course of GVD disease following ex vivo perfusion prior to transplantation of the donor intestine with the MRC OX19-ricin A chain conjugate.