Background: Immunotherapy utilizing T cells genetically modified to express chimeric antigen receptors (CARs) is rapidly emerging as a promising novel treatment for hematological and nonhematological malignancies. In order to target the TKI-insensitive leukemia stem cells (LSCs) in chronic myeloid leukemia (CML) by CAR T cells, we chose CD26 as a cell surface tumor-associated antigen due to preferentially expression on LSCs. Additionally, CD26 has also been suggested to be a multipurpose therapeutic target for other cancer. Therefore, developing the CD26-targeting CAR T cells may be a promising therapy for not only LSCs but also other CD26+ cancer cells. Methods: We designed the second-generation CD26-targeting CAR utilizing 4-1BB (CD137) as costimulatory domain, and transduced T cells with CD26-CAR containing lentiviral. Then we evaluated the transduction efficiency and expansion ability, and demonstrated the existence of self-antigen-driven fratricide by cytokine assay and cytotoxicity assay. Results: Anti-CD26-4-1BB-CAR T cells exhibited poor viability, multiple cytokine secretion, down-regulation of CD26 and direct cytotoxicity against themselves, indicating self-antigen-driven fratricide. Conclusion: Eradicating CML-LSCs via anti-CD26-4-1BB-CAR T cells is not applicable, and optimized design or alternative target is needed.