In the clinical daily life various lesions of the oral cavity have shown different aspects, generating an inconclusive or doubtful diagnosis. In general, oral injuries are diagnosed by histopathological analysis from biopsy, which is an invasive procedure and does not gives immediate results. In the other hand, Raman spectroscopy technique it is a real time and minimal invasive analytical tool, with notable diagnostic capability. This study aims to characterize, by optical fiber Raman-based spectroscopy (OFRS), normal, inflammatory, potentially malignant, benign and malign oral lesions. Raman data were collected by a Holospec f / 1.8 spectrograph (Kayser Optical Systems) coupled to an optical fiber, with a 785nm laser line source and a CCD Detector. The data were pre-processed and vector normalized. The average analysis and standard deviation was performed associated with cluster analysis and compared to the histopalogical results. Samples of described oral pathological processes were used in the study. The OFRS was efficient to characterized oral lesions and normal mucosa, in which biochemical information related to vibrational modes of proteins, lipids, nucleic acids and carbohydrates were observed. The technique (OFRS) is able to demonstrate biochemical information concern different types of oral lesions showing that Raman spectroscopy could be useful for an early and minimal invasive diagnosis.