Chimeric antigen receptor (CAR)-T therapy on acute myeloid leukemia (AML) is hindered by the absence of a suitable tumor-specific antigen. Here, we propose CD97 as a potential target for CAR-T therapy against AML based on its broader and higher expression on AML cells compared to normal hematopoietic stem and progenitor cells (HSPCs). To resolve the fratricide problem caused by CD97 expression on T cells, we knock out CD97 in CAR-T cells using CRISPR-Cas9. Our CD97 CAR-T cells eliminate both AML cell lines and primary AML cells effectively while showing tolerable toxicity to HSPCs. Furthermore, we mutate the CD3ζ domain of the CAR and find that the optimized CD97 CAR-T cells exhibit persistent anti-tumor activity both in vitro and in multiple xenograft models. Mechanistically, transcriptional profiles reveal that the optimized CAR-T cells delay differentiation and resist exhaustion. Collectively, our study supports CD97 as a promising target for CAR-T therapy against AML.