Background We have demonstrated previously that retroviral-mediated transfer of a suicide gene into bone marrow (BM) donor T cells allows an efficient control of graft-versus-host disease (GvHD) after allogeneic BM transplantation. However, the 12 days of ex vivo culture required for the production of gene-modified cells (GMC), including soluble CD3 monoclonal antibody (MAb)-mediated activation and expansion with interleukin (IL)-2, induced a decrease of GMC alloreactivity and a reversal of their CD4/CD8 ratio. Improving the culture protocol in order to maintain the highest alloreactivity is of critical importance in obtaining an optimal graft-versus-leukemia (GvL) effect. Methods Peripheral blood mononuclear cells were activated with soluble CD3 MAb or CD3 and CD28 MAb co-immobilized on beads and expanded for 12 days in the presence of IL-2, IL-7 or IL-15 before analysis of alloreactivity and phenotype. Results Replacing the CD3 MAb by CD3/CD28 beads led to similar in vitro alloreactivity but improved the expansion and in vivo alloreactivity of GMC. Replacing the IL-2 with IL-7, but not IL-15, or decreasing IL-2 or IL-7 concentrations, improved the in vitro alloreactivity of expanded cells but was associated with lower expansion. Indeed, the alloreactivity of expanded cells was negatively correlated with cell expansion and positively correlated with CD4/CD8 ratio and CD8 expression level. Discussion Quantitative (i.e. low CD4/CD8 ratio) and qualitative (e.g. low CD8 expression) defects may account for the decreased alloreactivity of GMC. Using CD3/CD28 beads and/or IL-7 is more beneficial than CD3 MAb and IL-2 for preventing perturbations of the alloreactivity and phenotype of GMC.