The indubitable existence of a graft-versus-lymphoma (GVL) effect is difficult to prove directly. This article reviews the difficulties in interpreting the current literature in this field and, with a number of caveats, argues for the existence of a clinically meaningful GVL effect in follicular, mantle cell, small lymphocytic, and Hodgkin lymphomas. The evidence, however, for a potent GVL effect in diffuse large-cell lymphoma and Burkitt lymphoma is not convincing. Policies for allografting in lymphoma are proposed on the basis of this evidence. The immunobiology of GVL effects is discussed--in particular, the expression of HLA class I and II and co-stimulatory molecules on lymphomas that influence the generation of alloreactive T cells--together with future directions in immunotherapy that may help to eradicate chemoresistant disease.