In this paper, light-emitting diode (LED) based two-dimensional fluorescence correlation spectroscopy was used to discriminate tea leaves with different grades. The distance between LED and tea samples was changed as an external variable. As the fluorescence spectral data collected through the experiment was large, principal component regression (PCR) was used to extract the important information and analyze the spectral data. The final two-dimensional fluorescence correlation spectra contour maps showed obvious difference between different tea leaves and the predictive results based on the leave-one-out method. It showed the strong ability of this spectral method for tea classification.