Acute myeloid leukemia (AML) is an aggressive and genetically heterogeneous disease with poor clinical outcomes. Refractory AML is common, and relapse remains a major challenge, attributable to the presence of therapy-resistant leukemic stem cells (LSCs), which possess self-renewal and repopulating capability. Targeting LSCs is currently the most promising avenue for long-term management of AML. Likewise, chimeric antigen receptor (CAR)-natural killer (NK) cells have emerged as a promising alternative to CAR-T cells due to their intrinsic potential as off-the-shelf products and safer clinical profiles. Here, we introduced a third-generation CAR harboring TIM3 scFv, CD28, 4-1BB, and CD3ζ (CAR-TIM3) into human NK-92 cells, the only FDA-approved NK cell line for clinical trials. TIM3 was chosen as a target antigen owing to its differential expression in LSCs and normal hematopoietic stem/progenitor cells (HSPCs). The established CAR-TIM3 NK-92 cells effectively targeted TIM3 and displayed potent anti-tumor activity against various primitive AML cells, subsequently causing a reduction in leukemic clonogenic growth in vitro, while having minimal effects on HSPCs. CAR-TIM3 NK-92 cells significantly reduced leukemic burden in vivo and interestingly suppressed the engraftment of AML cells into the mouse liver and bone marrow. Surprisingly, we found that CAR-TIM3 NK-92 cells expressed relatively low surface TIM3, leading to a low fratricidal effect. As TIM3 and PD-1 are immune checkpoints involved in NK cell dysfunction, we further tested and found that CAR-TIM3 NK-92 cells are beneficial for alleviating NK cell exhaustion. Our findings highlight the potential application of CAR-TIM3 NK cells for cellular immunotherapy for TIM3+ AML.