Background: Approximately fifty percent of patients with acute myeloid leukemia can be cured with current therapeutic strategies which include, standard dose chemotherapy for patients at standard risk of relapse as assessed by cytogenetic and molecular analysis, or high-dose chemotherapy with allogeneic hematopoietic stem cell transplant for high-risk patients. Despite allogeneic hematopoietic stem cell transplant about 25% of patients still succumb to disease relapse, therefore, novel strategies are needed to improve the outcome of patients with acute myeloid leukemia. Methods and findings: We developed an immunotherapeutic strategy targeting the CD33 myeloid antigen, expressed in ∼ 85-90% of patients with acute myeloid leukemia, using chimeric antigen receptor redirected T-cells. Considering that administration of CAR T-cells has been associated with cytokine release syndrome and other potential off-tumor effects in patients, safety measures were here investigated and reported. We genetically modified human activated Tcells from healthy donors or patients with acute myeloid leukemia with retroviral supernatant encoding the inducible Caspase9 suicide gene, a ÄCD19 selectable marker, and a humanized third generation chimeric antigen receptor recognizing human CD33. ÄCD19 selected inducible Caspase9-CAR.CD33 T-cells had a 75±3.8% (average ± standard error of the mean) chimeric antigen receptor expression, were able to specifically lyse CD33+ targets in vitro, including freshly isolated leukemic blasts from patients, produce significant amount of tumor-necrosis-factor-alpha and interferon-gamma, express the CD107a degranulation marker, and proliferate upon antigen specific stimulation. Challenging ÄCD19 selected inducible Caspase9-CAR.CD33 T-cells with programmed-death-ligand-1 enriched leukemia blasts resulted in significant killing like observed for the programmed-death-ligand-1 negative leukemic blasts fraction. Since the administration of 10 nanomolar of a non-therapeutic dimerizer to activate the suicide gene resulted in the elimination of only 76.4±2.0% gene modified cells in vitro, we found that co-administration of the dimerizer with either the BCL-2 inhibitor ABT-199, the pan-BCL inhibitor ABT-737, or mafosfamide, resulted in an additive effect up to complete cell elimination. Conclusions: This strategy could be investigated for the safety of CAR T-cell applications, and targeting CD33 could be used as a "bridge" therapy for patients coming to allogeneic hematopoietic stem cell transplant, as anti-leukemia activity from infusing CAR.CD33 T-cells has been demonstrated in an ongoing clinical trial. Albeit never performed in the clinical setting, our future plan is to investigate the utility of iC9-CAR.CD33 T-cells as part of the conditioning therapy for an allogeneic hematopoietic stem cell transplant for acute myeloid leukemia, together with other myelosuppressive agents, whilst the activation of the inducible Caspase9 suicide gene would grant elimination of the infused gene modified T-cells prior to stem cell infusion to reduce the risk of engraftment failure as the CD33 is also expressed on a proportion of the donor stem cell graft.