This work introduces a dual-excitation, multispectral-detection, frequency-domain fluorescence lifetime imaging (FLIM) handheld endoscopy system for comprehensive autofluorescence imaging of the oral mucosa. The excitation wavelengths and spectral detection channels were selected to preferentially image the fluorescence of endogenous fluorophores relevant for oral epithelial cancer detection. The endoscope dimensions and optical specifications enable direct imaging of oral lesions with minimal patient discomfort. Comprehensive imaging experiments using standard fluorophores demonstrate the capabilities of the FLIM endoscope to measure a wide range of fluorescence lifetimes (0.1-12 ns) with high precision (∼100 ps). Clinical in vivo label-free metabolic and biochemical endoscopic FLIM imaging was demonstrated by imaging a malignant oral lesion and healthy mucosa in a patient undergoing a tissue biopsy procedure.