When treatment fails, the clinical outcome of acute leukemia patients is usually very poor, particularly when failure occurs after transplantation. A second allogeneic stem cell transplant could be envisaged as an effective and feasible salvage option in younger patients having a late relapse and an available donor. Unmanipulated or minimally manipulated donor T cells may also be effective in a minority of patients but the main limit remains the induction of severe graft-versus-host disease. This clinical complication has brought about a huge research effort that led to the development of leukemia-specific T-cell therapy aiming at the direct recognition of leukemia-specific rather than minor histocompatibility antigens. Despite a great scientific interest, the clinical feasibility of such an approach has proven to be quite problematic. To overcome this limitation, more research has moved toward the choice of targeting commonly expressed hematopoietic specific antigens by the genetic modification of unselected T cells. The best example of this is represented by the anti-CD19 chimeric antigen receptor (CD19.CAR) T cells. As a possible alternative to the genetic manipulation of unselected T cells, specific T-cell subpopulations with in vivo favorable homing and long-term survival properties have been genetically modified by CAR molecules. Finally, the use of naturally cytotoxic effector cells such as natural killer and cytokine-induced killer cells has been proposed in several clinical trials. The clinical development of these latter cells could also be further expanded by additional genetic modifications using the CAR technology.