Transplantation of hematopoietic stem cells from mismatched related donors makes a potential donor available for every child in need of stem cell transplantation. Here, we compare three different graft manipulation methods in patients with leukemias and lymphomas: positive selection of stem cells with either CD34 (n = 39) or CD133-coated magnetic microbeads (n = 14) and a new strategy which depletes T- and B-cells through the use of CD3- and CD19-coated microbeads (n = 11). Median purity of stem cells was comparable after CD34 (+)-selection and CD133 (+)-selection, whereas stem cells were only slightly enriched after CD3 (+)/CD19 (+)-depletion (97.5 %, 93.4 % and 1.02 %). Indirect depletion of T-cells by positive selection resulted in 1 x 10 (4) (median) residual CD3 (+)-cells/kg (0.7-3 x 10 (4)). Patients with CD3/CD19-depleted grafts received 3.2 x 10 (4) (median) (0.7-16 x 10 (4)) residual T-cells/kg. Those grafts also comprised NK-cells (median number: 86 x 10 (6)/kg), dendritic cells and monocytes/granulocytes. Primary engraftment of the stem cell products was comparable after CD34- and CD133-selection (85 and 72 %). In the CD3/CD19 group, 91 % had a primary engraftment. After reconditioning, all patients (64/64) were finally engrafted. Patients with CD34-selected or CD133-selected grafts had similar incidences of a GvHD II-IV (3 and 7 %), whereas a GvHD was slightly increased in patients receiving CD3/CD19-depleted cells (27 %). Reconstitution of CD3 (+) T-cells was faster in the CD3/CD19 group than in the CD34 or CD133 group. These preliminary results indicate, that CD3/CD19-selected grafts may be advantageous regarding engraftment and immunoreconstitution. Since effector cell with potential antileukemic activity are cotransfused, such grafts may be suited in particular for patients with insufficient remission.