a Although formalin fixed paraffin preserved (FFPP) tissues are a major resource for retrospective studies of disease progression, their use in vibrational spectroscopy studies has been undermined by issues of contributions of substrate and paraffin wax which persist in the spectra and can compromise spectral analysis. Recognising the microcrystalline nature of the wax in the tissue, which are inhomogeneously oriented with respect to the polarisation of the Raman source laser, in this study, we have developed a novel method for removing the paraffin wax contributions to the spectra using matrices of multiple wax spectra. FFPP tissue sections from the oral mucosca were obtained and, with no further chemical processing, the Raman spectral analysis of two regions, epithelium and connective tissue were compared. Matrices of multiple wax spectra were collected from different regions and subtracted from the epithelial and connective tissue spectra using a least squares analysis with non-negative constraints. Spectra of multiple cell components such as DNA and RNA were used in fitting the least squares model to reduce the residual error. The use of a data matrix of multiple wax spectra, as opposed to a single spectrum, results in a more accurate removal of the wax, hence reducing its contribution to spectral analysis. In unprocessed FFPP tissue sections, the contribution of the glass substrate is seen to be minimised in comparison to chemically dewaxed FFPP tissue sections. Contributions of the glass substrate were also successfully removed digitally using the same methodology. The combined results indicate that direct analysis of FFPP tissue sections is feasible using Raman spectroscopy, avoiding the need for chemical dewaxing. Additionally, the ability to use glass slides is very important in translation to the clinic.