Graft-versus-host disease (GVHD) remains the major problem to be overcome in transplantation of allogeneic haemopoietic stem cells. Using immunosuppressive prophylaxis with cyclosporin and methotrexate, moderate to severe acute GVHD develops in approximately 45% of transplant recipients with an HLA-identical sibling donor and in >75% of patients from unrelated HLA-identical or partially matched related donors. The pathophysiology of GVHD is complex and still incompletely described. Experimental and clinical data indicate that GVHD is largely mediated by immunocompetent T cells in the donor stem cell graft which are reactive against recipient (host) tissues. Depletion of these immunocompetent T cells from the stem cell graft offers a way to effectively prevent GVHD. The first section of this review describes the technical principles of different methods of T cell depletion. The advantages, limitations and level of T cell depletion achievable by physical methods or by positive and negative immunoselection procedures using monoclonal antibodies are comprehensively discussed. A short section concentrates on technical problems in the enumeration of T cells in the context of depletion efficiency. In the section on clinical studies, the focus is on the efficacy of different T cell depletion methods in avoiding GVHD in different clinical settings. The various methods are compared in transplantation from HLA-identical and nonidentical siblings or matched unrelated donors. The major drawbacks of T cell depletion are discussed in detail. Failure of engraftment and graft rejection is a more frequent problem following T cell-depleted transplants, particularly with HLA nonidentical donor-recipient pairs. An increase in leukaemic relapse rate is seen in certain haematological malignancies, especially in chronic myeloid leukaemia. Delayed recovery of anti-infectious immunity occurs, leading to an increased incidence of cytomegalovirus and Epstein-Barr virus related problems. The aim of this review is not only to give an overview of published studies but also to review strategies to circumvent the drawbacks of TCD. Consequently, we attempt to describe the potential role of cells removed by different depletion methods in graft protection, anti-infectious immunity and graft-versus-leukaemia reactivity. Finally, the possibility of recovering all components of the original graft and readministering them in controlled amounts and at controlled times is discussed. This strategy of ‘balanced component therapy’ may allow the combination of a low rate and severity of GVHD without the disadvantages of T cell depletion.