Purpose: Although molecular analyses have been shown to be useful for classifying tumors and formulating a prognosis for patients, current methods applied to glioma are mostly based on histopathology. To search for robust diagnostic and prognostic markers of high-grade gliomas (HGGs), we applied a meta-analysis that extracts the most reliable information from the available gene expression data. Methods: We obtained data sets from three studies in HGGs. We selected a consensus set of 267 patients. Differential analyses were performed separately for each study and we applied a meta-analysis approach based on nonparametric rank products to evaluate the combined data. The genes selected in both analyses were used to construct a gene classifier by logistic regression modeling. We performed survival analyses on 144 patients by fitting Cox proportional hazard model. Genes identified as both differentially expressed and correlated to survival were used to build an optimal survival model. Performances were evaluated on an independent data set comprising 54 patients. Several genes were validated using RT, Q-PCR, and immunohistochemistry on a local cohort of 130 patients. Results: This interstudy cross-validation approach generated a set of 65 genes consistently and specifically differentially expressed in GBM. Functional annotation revealed a clear association with the nervous system development and the cell communication. The genes significantly associated with grading were mostly related to the extracellular matrix. The optimal survival model was built on a four genes signature. Kaplan-Meier curves and the log rank test (training: p52 e-11, testing: p51 e-4) indicated the high survival prognostic potential for this classifier. Finally, applied only to GBM it clearly outperformed previous reports, by grouping GBM in two subtypes with significant different prognosis (p53 e-4). Conclusions: Meta-analysis allows the identification and validation of HGGs biomarkers that might represent good candidates for novel diagnostic and prognostic approaches in HGGs.