Introduction: The aim of the study was to construct a pyroptosis-related risk score (RS) model for the prognosis of acute myeloid leukemia (AML). Methods: The TARGET (training) and E-MTAB-1216 (validation) datasets were downloaded. Pyroptosis-related genes with differences in expression were identified between the recurrent and nonrecurrent samples of the training dataset. An RS prognostic model comprising seven pyroptosis-related genes was constructed using LASSO regression coefficients. The samples were classified into the high- and low-risk groups using the RS model; the differentially expressed genes (DEGs) between these groups were identified, followed by DEG functional analysis and the immunological evaluation of these groups. Results: Forty-nine pyroptosis-related genes, including 22 DEGs, were screened. WT1, NPM, FLT3/ITD, and CEBPA mutations were found in most pediatric AML samples. An RS prognostic model was constructed using 7 pyroptosis-related genes. The two risk groups and prognostic data were significantly related. FLT3/ITD mutations, CEBPA mutations, and RS model status were identified as independent prognostic factors, using the clinical information. The DEGs between the two groups were correlated with immune-related pathways. Moreover, the immune cell distribution and the occurrence of immune-related pathways were notably decreased in the high-risk group. Discussion/Conclusion: Seven pyroptosis-related genes, CHMP2A, PRKACA, CASP9, IRF2, CHMP3, HMGB1, and AIM2, can predict the prognosis and recurrence of childhood AML.