Oral cancer is a public health problem with relevant incidence in the world population. The affected patient usually presents advanced stage disease and the consequence of this delay is a reduction in survival rates. Given this, it is essential to detect oral cancer at early stages. Fluorescence spectroscopy is a non-invasive diagnostic tool that can improve cancer detection in real time. It is a fast and accurate technique, relatively simple, which evaluates the biochemical composition and structure using the tissue fluorescence spectrum as interrogation data. Several studies have positive data regarding the tools for differentiating between normal mucosa and cancer, but the difference between cancer and potentially malignant disorders is not clear. The aim of this study was to evaluate the efficacy of fluorescence spectroscopy in the discrimination of normal oral mucosa, oral cancer, and potentially malignant disorders. The fluorescence spectroscopy was evaluated in 115 individuals, of whom 55 patients presented oral squamous cell carcinoma, 30 volunteers showing normal oral mucosa, and 30 patients having potentially malignant disorders. The spectra were classified and compared to histopathology to evaluate the efficiency in diagnostic discrimination employing fluorescence. In order to classify the spectra, a decision tree algorithm (C4.5) was applied. Despite of the high variance observed in spectral data, the specificity and sensitivity obtained were 93.8% and 88.5%, respectively at 406 nm excitation. These results point to the potential use of fluorescence spectroscopy as an important tool for oral cancer diagnosis and potentially malignant disorders.