Oral cancer is the most common cancer among Indian males, with 5-year- survival-rates of less than 50%. Efficacy of Raman spectroscopic methods in non-invasive and objective diagnosis of oral cancers and confounding factors has already been demonstrated. The present Raman microspectroscopic study was undertaken for in-depth and site-specific analysis of normal and tumor tissues. 10 normal and 10 tumors unstained sections from 20 tissues were accrued. Raman data of 160 x 60 μm and 140 x 140 μm in normal and tumor sections, respectively, were acquired using WITec alpha 300R equipped with 532 nm laser, 50X objective and 600 gr/mm grating. Spectral data were corrected for CCDresponse, background. First-derivitized and vector-normalized data were then subjected to K-mean cluster analysis to generate Raman maps and correlated with their respective histopathology. In normal sections, stratification among epithelial layers i.e. basal, intermediate, superficial was observed. Tumor, stromal and inflammatory regions were identified in case of tumor section. Extracted spectra of the pathologically annotated regions were subjected to Principal component analysis. Findings suggest that all three layers of normal epithelium can be differentiated against tumor cells. In epithelium, basal and superficial layers can be separated while intermediate layer show misclassifications. In tumors, discrimination of inflammatory regions from tumor cells and tumor-stroma regions were observed. Finding of the study indicate Raman mapping can lead to molecular level insights of normal and pathological states.