Background: Changes in the bone marrow microenvironment are believed to play a major role in the biology of Myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML). To study the bone marrow microenvironment (BME) in MDS and AML and to compare it with normal BME, we studied the expression profile of 43 immune biomarkers and evaluated the differences in the BME between AML and MDS and compared it to that of normal BME. These 43 immune biomarkers included B- and T-cell markers, cytokines and chemokines. Methods: RNA was extracted from fresh bone marrow aspiration samples from 626 patients with AML, 564 patients with MDS, and 1449 individuals having bone marrow without any mutations or having low level mutations determined to be CHIP (clonal hematopoiesis of indeterminate potential) and considered normal. RNA levels of 42 immune biomarkers were quantified using next generation sequencing. Using a machine learning algorithm, we first selected the relative genes that distinguish between two classes using K-fold cross-validation (K = 12). The selected genes were used to predict one class from the other using random forest classifier. Samples were divided into a training set (67%) and testing set (33%) for each classification. Results: The random forest showed that MDS can easily be distinguished from normal using the expression of 15 genes (CYFIP2, CXCR4, IL1RAP, CD58, CD36, CD19, PAX5, CD79B, ID1, IL8, CD44, IL1R1, CD79A, IL21R, and CD74). The AUC for distinguishing MDS from normal was 0.996 in the training set and 0.931 (95% CI: 0.912-0.949) in the testing set. Distinguishing between AML and normal was also robust and achievable using the expression of only 10 genes (CYFIP2, IL1R1, CXCR4, IL8, IL21R, CD44, CD28, CD79A, and IL7R, and CD8A). The AUC for the training set was 0.994 and 0.972 (95% CI: 0.961-0.983) for the testing set. Eight of these 10 markers were shared with MDS algorithm. Only CD28 and IL7R were specifically needed for AML classification. Distinguishing between MDS And AML was achievable with high reliability with AUC of 0.994 (95% CI:0.992-0.997) in training set and 0.924 (95% CI: 0.896-0.952) in testing set). Only 10 biomarkers were used for distinguishing MDS from AML, nine of which (IL1R1, CYFIP2, CD44, IL1RAP, CXCR4, IL21R, CD74, IL8, and CD36) were used in distinguishing MDS from normal. The only unique biomarker was CD28. Comparing levels of these biomarkers, most of which showed highest level in normal BM, significantly lower level in MDS but the level was even significantly lower in AML (deeper reduction) than in MDS. Conclusions: This data suggests that the BME is significantly different in MDS from AML and both are different from normal. Few immune biomarkers play major role in defining each BME. However, relative increase or decrease between these immune biomarkers dictate the uniqueness of each microenvironment.