Background <p>Glioblastoma multiforme (GBM) is an aggressive brain tumor with poor prognosis. Microglia are key immune cells in the GBM tumor microenvironment (TME), yet their subtype-specific impact, particularly inflammatory microglia, remains poorly understood.</p> Methods <p>We analyzed single-cell RNA sequencing data to characterize inflammatory microglia patterns in GBM. Patient subgroups were compared for clinical features, molecular profiles, and predicted immunotherapy responses.</p> Results <p>Univariate Cox regression identified 35 inflammatory microglia-related genes associated with overall survival (OS). Based on these genes, patients were stratified into two reproducible subtypes (Cluster1 and Cluster2) across three independent datasets. Cluster2 displayed higher OS and better predicted immunotherapy response. It showed lower inflammatory microglia gene expression, reduced cytokine/chemokine activity, enhanced neural pathway activity, lower stromal and immune infiltration, decreased epithelial-mesenchymal transition, higher tumor purity, increased genomic instability—including <i>IDH1</i>, <i>ATRX</i>, and <i>TP53</i> mutations—and lower <i>MGMT</i> methylation. Cluster1 exhibited the opposite pattern, with higher inflammatory gene expression, an immunosuppressive TME, and elevated MGMT methylation.</p> Conclusion <p>Inflammatory microglia-based stratification reveals key biological heterogeneity in GBM and may guide prognosis and therapeutic decision-making.</p>