Gliomas are characterized by their highly heterogeneous aggressiveness even in the same histological category. The biological features of gliomas would be precisely described by global gene expression data at protein level. We examined the proteome of 93 surgical samples including 52 glioblastomas, 13 anaplastic astrocytomas, 10 astrocytomas, 8 oligodendrogliomas, and 10 normal brain tissue samples by using two-dimensional gel electrophoresis and mass spectrometry. On the hierarchical clustering developed from the proteome profiling patterns, the normal brain tissues and oligodendrogliomas were identified as distinct subsets. Furthermore, the proteome-based clustering significantly correlated with the patients'survival, and we could identify a biologically distinct subset of astrocytoma with aggressive nature. The discriminant analysis extracted a set of 37 proteins differentially expressed by the histological grading. Then, to identify the protein markers that are clinically useful for predicting efficacy of anticancer agents, we also investigated the correlation between the proteome profiling patterns and the in-vitro chemosensitivity measured by flow cytometric detection of apoptosis. The clustering map provided a particular distribution of chemosensitivity spectrum for each sample. We identified a set of 41 proteins that significantly affected the in-vitro chemosensitivity by discriminant analysis based on the proteome data. Many of the proteins that correlated with chemoresistance were categorized into the signal transduction proteins including the small G-proteins. The present study has shown that the proteome analysis is useful to develop a novel system for the prediction of biological aggressiveness and chemosensitivity. The proteins identified here could be novel biomarkers for survival prediction and rational targets for anti-glioma therapy.