As an optical spectroscopic technique, Raman spectroscopy (RS) has shown promise for in vivo cancer diagnostics in various organs. Common Raman signals are extremely low and easily disturbed by fluorescence. In this study, we used surface-enhanced RS to scan patients’ malignant and adjacent tissues, as well as the normal tissues. Raman scattering signals were acquired from 35 normal tissue samples and 39 laryngeal carcinoma tissue samples. Linear discriminant analysis (LDA), based on principal component analysis, was employed to generate diagnostic algorithms for the classification of different laryngeal tissue types. We found there were several variations in surface-enhanced RS (SERS) peaks, and the same with other malignant ones. When using the principle component-LDA (PC–LDA) algorithms, this study showed that the diagnostic sensitivity and specificity were 99.2% and 98.4%, respectively for the identifying cancer tissues from normal samples. Results show that there are significant spectral differences (different peak intensities) between cancer tissues and normal tissues. High classification accuracy can be achieved by using the PC–LDA diagnostic algorithm, which demonstrates the great potential of tissue-SERS analysis combined with PC–LDA diagnostic algorithms for cancer screening in laryngeal cancer.