A new method for fast diagnosis and mapping of tooth caries is suggested. The method is based on autofluorescence and application of Fourier transform imaging microscopy. This device provides simultaneous fluorescence spectra at all pixels of the examined sample. A cooled CCD detector, directly connected to the spectrometer, collects the imaging interferograms and allows for spectral mapping. The feasibility of this method was evaluated by examining human teeth for caries detection. Images and the corresponding spectra were acquired at various tooth regions. It was found that a distinct spectral feature characterizes the carious regions. Spectral analysis allows for detection of even early stages of caries, while the associated mapping capability allows for morphological characterization. The spatial resolution depends on the specific optical setup used, and in our case it was ca. 1 μm. These findings suggest that this technique may be further developed and applied for clinical detection of initial and well established caries during oral examination and diagnosis.