Since its first emergence as an unexpected and poorly understood phenomenon in experimental animals, GVHD has become much better characterised both clinically and biologically. Unfortunately prevention of GVHD has only been achieved at the price of an increased risk of graft rejection and leukaemic relapse and treatment of the established syndrome is only effective in a proportion of patients. It is to be hoped that the more detailed picture of the process now emerging will facilitate better immune manipulation of the bone marrow transplant so as to select recovery of only the beneficial aspects of donor immunity namely immunity against microorganisms and graft versus leukaemia reactivity.