Using counterflow centrifugation elutriation (CCE) lymphocytes can be separated from CD34+ populations based on size. Immature progenitors tend to be smaller than mature cells suggesting that CCE introduces loss of stem cells. We compared the separation of 12 PBSC with 16 BM transplants. Cells were separated in 12 fractions (3000-2200 r.p.m.) and the rotor off (RO) fraction. Separation patterns of BM and PBSC were comparable. B cells were collected in the high speed fractions followed by T and NK cells. In contrast, progenitor cells were collected in lower speed fractions. By adding successively T cell-depleted fractions to the RO fraction a BM transplant could be composed containing 0.7 x 10(6) T cells/kg and 90%, 89% and 68% recovery of CD34+, CFU-GM and BFU-E. PBSC were separated in four CCE runs inducing higher numbers of T cells in the graft (4.4 x 10(6)/kg) and 54% CD34+, 46% CFU-GM and 37% BFU-E recovery. Time of engraftment was not delayed and no graft failure was observed. The higher number of T cells was not associated with higher incidence of GVHD. Acute GVHD > or = grade III occurred in 0 of 16 BM and two of 12 PBSC recipients; extensive chronic GVHD was observed in four of 15 and three of nine recipients, respectively. To study immature cells in the graft, CD34 subpopulations and cells with long-term repopulating ability, determined using cobble-stone area formation (CAFC assay), were evaluated in each fraction. The separation patterns in BM and PBSC were comparable. Cells with mature and immature phenotype were enriched in lower speed fractions (mean recovery of 74% CD34+/CD13-/DR-). The CAFC week 2, 4 and 6 were also enriched in these fractions. These data show that the used CCE procedure is a reliable method to deplete T cells from stem cell transplants without substantial loss of immature and mature progenitors.