Background: This study mainly used The Cancer Genome Atlas (TCGA) RNA sequencing dataset to screen prognostic snoRNAs of acute myeloid leukemia (AML), and used for the construction of prognostic snoRNAs signature for AML. Methods: A total of 130 AML patients with RNA sequencing dataset were used for prognostic snoRNAs screenning. SnoRNAs co-expressed genes and differentially expressed genes (DEGs) were used for functional annotation, as well as gene set enrichment analysis (GSEA). Connectivity Map (CMap) also used for potential targeted drugs screening. Results:Through genome-wide screening, we identified 30 snoRNAs that were significantly associated with the prognosis of AML. Then we used the step function to screen a prognostic signature composed of 14 snoRNAs (SNORD72, SNORD38, U3, SNORA73B, SNORD79, SNORA73, SNORD12B, SNORA74, SNORD116-12, SNORA65, SNORA14, snoU13, SNORA75, SNORA31), which can significantly divide AML patients into high- and low-risk groups. Through GSEA, snoRNAs co-expressed genes and DEGs functional enrichment analysis, we screened a large number of potential functional mechanisms of this prognostic signature in AML, such as phosphatidylinositol 3-kinase-Akt ,Wnt, epithelial to mesenchymal transition, T cell receptors, NF-kappa B, mTOR and other classic cancer-related signaling pathways. In the subsequent targeted drug screening using CMap, we also identified six drugs that can be used for AML targeted therapy, they were alimemazine, MG-262, fluoxetine, quipazine, naltrexone and oxybenzone. Conclusion: Our current study was constructed an AML prognostic signature based on the 14 prognostic snoRNAs, which may serve as a novel prognostic biomarker for AML.