We here use knockin mutagenesis in the mouse to model the spectrum of acquired CEBPA mutations in human acute myeloid leukemia. We find that C-terminal C/EBPα mutations increase the proliferation of long-term hematopoietic stem cells (LT-HSCs) in a cell-intrinsic manner and override normal HSC homeostasis, leading to expansion of premalignant HSCs. However, such mutations impair myeloid programming of HSCs and block myeloid lineage commitment when homozygous. In contrast, N-terminal C/EBPα mutations are silent with regards to HSC expansion, but allow the formation of committed myeloid progenitors, the templates for leukemia-initiating cells. The combination of N- and C-terminal C/EBPα mutations incorporates both features, accelerating disease development and explaining the clinical prevalence of this configuration of CEBPA mutations.