In order to build a prognostic risk model for patients with acute myeloid leukemia (AML), this study extracted immune genes through IMMPORT database, the Gene Expression Omnibus (GEO) database was used to search and download AML-related transcriptome RNA sequencing datasets, and the differentially expressed immune genes in AML were searched and analyzed. And we downloaded the AML related expression profile dataset (TCGA-LAML) and clinical data based on the Cancer Genome Atlas (TCGA) database, the differentially expressed immune genes were further analyzed by constructing the prognostic risk model through univariate COX regression analysis and multivariate COX regression analysis. The key immune genes were finally determined, and the single gene prognosis analysis and immune infiltration analysis were carried out according to the immune genes. The results of analysis showed that compared with the normal control group, 55 immune genes were significantly up-regulated or down-regulated in the bone marrow samples of AML patients. The risk model based on four immune genes (FGF13, GZMB, FLT3 and CRLF3) could accurately predict the prognosis of AML patients ［in which the area under the curve (AUC) = 0.741］. There were significant differences in the content of immune cells in high/low-risk groups. FGF13 and CRLF3 were low-risk genes, GZMB and FLT3 were high-risk genes. We constructed a risk model based on four immune genes (FGF13, GZMB, FLT3, CRLF3), which provides a theoretical basis for monitoring the prognosis and immune infiltration of AML patients.