Allogeneic transplantation of granulocyte colony-stimulating factor (G-CSF)-mobilized peripheral blood progenitor cells (PBPC) appears to be an attractive alternative to allogeneic bone marrow transplantation (BMT). However, because vast amounts of potentially graft-vs.-host-reactive T cells are transfused with PBPC grafts, the use of PBPC in the allogeneic setting may be associated with an increased incidence or severity of graft-vs.-host disease (GVHD). To evaluate strategies for prevention of GVHD after PBPC allografting, we have studied T cell depletion (TCD) of G-CSF-mobilized PBPC samples harvested from six healthy donors and from five patients scheduled for autologous PBPC transplantation. Three approaches (CAMPATH-1 plus autologous complement [C], immunomagnetic CD34+ cell selection, and biotin-avidin-mediated CD34+ cell selection) were compared. TCD of PBPC samples with the monoclonal antibody (MAb) CAMPATH-1 plus autologous C resulted in a median elimination of 2.16 log CD3+ T cells, whereas 39% of CD56+ natural killer (NK) cells and 56% of CD34+ progenitor cells were recovered. TCD by CD34+ cell selection with the Isolex (Baxter, Munich, Germany) or Ceprate (CellPro, Bothell, WA) devices achieved median depletions (Isolex vs. Ceprate) of 4.04 vs. 3.12 log T cells and > 5 vs. 3.27 log NK cells while allowing the recovery of 36 vs. 27% CD34+ cells. The median purity of CD34+ cells in the final product was 1.7 (CAMPATH-1), 94 (Isolex), and 65% (Ceprate). We conclude that all methods tested effectively deplete T cells from PBPC preparations harvested from healthy donors. Whereas immunomagnetic CD34+ selection is most effective in terms of elimination of T cells, the less intensive T and NK cell depletions achieved with CAMPATH-1 might be advantageous with regard to retaining engraftment potential and graft-vs.-leukemia (GVL) activity of PBPC allografts.