The prevention of graft-versus-host disease by T-lymphocyte depletion of allografts prior to bone marrow transplantation has resulted in an increase in graft failure/rejection and relapse of disease. Evidence for the selective roles of specific T-lymphocyte subsets in each of the engraftment, graft-versus-host disease, and disease relapse processes has been presented, but sufficient clinical verification to support any hypothesis in this regard is lacking. In this paper we describe a convenient and highly flexible clinical laboratory method for depletion of specific T-lymphocytes in controllable quantities by the use of select monoclonal antibodies. Almost 3 log10 removal of CD2+ and CD8+ cells without significant loss of hematopoietic progenitors (CFU-GM, BFU-E, and CD34+ cells) can be reproducibly achieved. This method, employing soybean agglutination and immunomagnetic beads, is potentially adaptable to depletion of any cell subset or any tumor cell for which unique cell-surface antigen characteristics have been defined. In addition, our protocol is equally suited to the positive selection of stem cells and hematopoietic progenitors bearing the CD34 surface antigen.