Prostate cancer (PC) is the most common cancer in men after non-melanoma skin cancer in the United Kingdom (Cancer Research UK, 2019). Current diagnostic methods (PSA, DRE, MRI & prostate biopsy) have limitations as these are unable to distinguish between low-risk cancers that do not need active treatment from cancers which are more likely to progress. In addition, prostate biopsy is invasive with potential side effects. There is an urgent need to identify new biomarkers for early diagnosis and prognostication in PC. Raman spectroscopy (RS) is an optical technique that utilises molecular-specific, inelastic scattering of light photons to interrogate biological samples. When laser light is incident on a biological sample, the photons from the laser light can interact with the intramolecular bonds present within the sample. The Raman spectrum is a direct function of the molecular composition of the tissue, providing a molecular fingerprint of the phenotypic expression of the cells and tissues, which can give good objective information regarding the pathological state of the biological sample under interrogation. We applied a technique of drop coating deposition Raman (DCDR) spectroscopy using both blood plasma and sera to see if a more accurate prediction of the presence and progression of prostate cancer could be achieved than PSA which would allow for blood sample triage of patients into at risk groups. 100 participants were recruited for this study (100 blood plasma and 100 serum samples). Secondly, 79 prostate tissue samples (from the same cohort) were interrogated with the aid of Raman micro-spectroscopy to ascertain if Raman spectroscopy can provide molecular fingerprint that can be utilised for real time in vivo analysis. Multivariate analysis of support vector machine (SVM) learning and linear discriminant analysis (LDA) were utilised differently to test the performance accuracy of the discriminant model for distinguishing between benign and malignant mean plasma spectra. SVM gave a better performance accuracy than LDA with sensitivity and specificity of 96% and 97% respectively and an area under the curve (AUC) of 0.98 as opposed to sensitivity and specificity of 51% and 80% respectively with AUC of 0.74 using LDA. Slightly lower performance accuracy was also observed when blood serum mean spectra analysis was compared with blood plasma mean spectra analysis for both machine learning algorithms (SVM & LDA). Tissue spectral analysis on the other hand recorded an overall accuracy of 80.8% and AUC of 0.82 with the SVM algorithm compared to performance accuracy of 75% and AUC of 0.77 with LDA algorithm (better performance noted with the SVM algorithm). The small sample size of 79 prostate biopsy tissues was responsible for the low sensitivity and specificity. Therefore, the tissues were insufficient to describe all the variances in each group as well as the variability of the gold standard technique. Conclusion: Raman spectroscopy could be a potentially useful technique in the management of Prostate Cancer in areas such as tissue diagnosis, assessment of surgical margin after radical prostatectomy, detection of metastasis, Prostate Cancer screening as well as monitoring and prognosticating patients with Prostate Cancer. However, more needs to be done to validate the approaches outlined in this thesis using prospective collection of new samples to test the classification models independently with sufficient statistical power. At this stage only the fluid-based models are likely to be large enough for this validation process.