Glioblastoma (GBM) is the most aggressive, malignant primary brain tumor, which has abundant tumor-infiltrating immune cells and stroma in the tumor microenvironment (TME). So far, the TME landscape of GBM has not been elucidated. GBM samples were retrieved from TCGA and GEO databases. We used ESTIMATE and CIBERSORT algorithms to calculate risk score associated with TME, and immune cell infiltration (ICI) score of each patient is calculated by PCA. GSEA analysis is explored for each subgroup. Finally, the patient prognosis in different ICI score subgroup is determined. Two ICI clusters are determined in 208 GBM patients, and 207 differentially expressed genes (DGEs) are found between ICI clusters. And then, two gene clusters were determined. Finally, we obtained ICI score for each patient using principal component analysis (PCA). Patients were divided into high and low ICI score subgroups by setting the median as cutoff. Through GSEA, we found ECM-receptor interaction, mTOR signaling pathway, pathways in cancer, TGF-beta signaling pathway, and other immunosuppressive pathway related genes in the low ICI score group. Furthermore, patients with high ICI score group have more better prognosis. Targeting the stroma in GBM may be an effective new therapeutic approach, and the ICI score is an effective potential prognostic classifier of GBM.