Abstract. Glioblastoma multiforme (GBM) is one of the most malignant human tumors, with a uniformly poor outcome. One obstacle in curing malignant brain tumors is the limitation of conventional light microscopy in detecting microscopic residual tumor in biopsy samples from the perimeter of the surgically resected tumor. We further refined the identification of GBM tumor tissue at the sub-cellular level, utilising the technique of Synchrotron, sourced mid-infrared (mid-IR) spectromicroscopy. Paired, thin (5 μm) cryosections of snap-frozen human GBM tumor samples removed at elective surgery were mounted on glass slides (hematoxylin and eosin-stained tissue section) and calcium fluoride (CaF 2 ) windows (unstained tissue section for transmission spectromicroscopy), respectively. Concordance of tumor bearing areas identified in the stained section with the unstained IR tissue section was confirmed by the pathologist of the study. Compared with molecular signatures obtained from normal control brain tissue, unique spectroscopic patterns were detected in GBM tumor samples from 6 patients. The identifying features of GBM were: i) high protein-to-lipid ratios (amide I+II/CH 2 symmetric stretch; amide I+II/CH 2 +CH 3 symmetric and asymmetric stretch), and ii) considerable enhancement of the intensities of characteristic peaks at 2,957 and 2,871 cm -1 representing CH 3 asymmetric and symmetric stretch, respectively. Spectral data sets were subjected to Ward's algorithm for assignment to similar groups, and then subjected to hierarchical cluster analysis (HCA) by means of false color digital maps. False color images of 5 clusters obtained by HCA identified dominant clusters corresponding to tumor tissue. Corroboration of these findings in a larger number of GBM may allow for more precise identification of these and other types of brain tumors.