Background: Acute myeloid leukemia (AML) is a common hematopoietic tumor with extremely high morbidity and mortality. This study was designed to explore candidate genes that were related to the poor prognosis of AML patients and analyze their relationship with drug sensitivity. Methods: Microarray databases were performed to screen the differentially expressed genes (DEGs). DAVID 6.8 was used for further functional enrichment analysis. The protein-protein interaction (PPI) network was constructed through STRING website and Cytoscape tool. Then, we analyzed and explored the mRNA transcription level, prognosis correlation, and drug sensitivity of the candidate genes in AML via multiple acknowledged databases including the GEPIA, BloodSpot, EMBL-EBI, UALCAN, LinkedOmics, and GSCALite databases. Results: A total of 181 up-regulated DEGs were screened. Three candidate genes (MAP2K3, LST1, and CYTH4) related to poor outcomes of AML patients were identified. Meanwhile, the high expression levels of the three genes were verified in AML patients and AML cell lines, the expression differences of three genes at AML different subtypes were demonstrated. Drug sensitivity analysis displayed the expression levels of MAP2K3, LST1, and CYTH4 were negatively related to drug resistance, indicating that the three genes were sensitive to certain small-molecule drugs (including targeted drugs and non-targeted drugs). Conclusion: In summary, MAP2K3, LST1, and CYTH4 may be potential prognostic indicators for AML, and may be associated with the sensitivity of certain small molecule drugs.