Treatment with hypomethylating agents (HMA), and specifically azacitidine (AZA), is the standard of care for patients with higher-risk myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML) that are not eligible to receive intensive chemotherapy. Despite research efforts, it is not possible to predict response to treatment with HMA. In this study, we aimed to identify immune cell signatures in the bone marrow (BM) associated with treatment outcomes. By employing mass cytometry, we performed an in-depth immunophenotypic analysis in BM samples deriving from patients with myeloid neoplasms prior to treatment initiation. We identified an increased pre-treatment frequency of a CD8+ T cell subset, characterized as CD57+CXCR3+CCR7-CD45RA+, in patients with MDS and AML who did not respond to treatment with AZA, compared to responders. Furthermore, an increased baseline frequency (>29%) of CD57+CXCR3+CD8+ T cells was correlated with poor overall survival. We further engaged scRNA-seq to assess the transcriptional profile of BM CD8+ T cells from treatment-naive patients with MDS and AML, to identify molecular signatures in CD8+ T subpopulations associated with favorable outcomes. Response to treatment was positively associated with enrichment of IFN-mediated pathways coupled with enhanced cytotoxic signature, whereas enrichment of the TGF-b signaling pathway was observed in cell clusters from non-responders. Together, this study identified a specific CD57+CXCR3+ CD8+ T cell population with predictive value in patients with MDS and AML treated with AZA and characterized molecular signatures in CD8+ T cells linked to cytokine signaling that were associated with treatment outcomes.