In this paper,LED based two-dimensional fluorescence correlation spectra were used to discriminate tea leaves with different grades.Generalized two-dimensional(2D)correlation spectra proposed by Noda in 1993 has become a useful technique to display subtle spectral changes induced by an external variable such as time,temperature and so on.In this paper,the distance between LED and tea samples was changed as an external variable.In the fluorescence system,LED with wavelength at 400nm placed above the sample was used as excitation source.The intensity of the LED was controlled through a data acquisition card(DAQ).A stepper motor connected with the LED holder was used to adjust the distance between the LED and tea samples.Both the DAQ and the stepper motor were controlled by a LabVive program automatically.Tea sample was placed on black anodized aluminium with negligible fluorescence.A long pass filter with cut-off wavelength at 450nm was used to filter out excitation light of the LED.Then only the induced fluorescence signal was collected by a multi-mode optical fiber and recorded by a spectrometer.The spectrometer integral time and LED-amplitude were fixed through the measurements.As the fluorescence spectra data collected through the experiment was large,principal component regression(PCR)was used to extract important information and analyze the spectra data.The final two-dimensional fluorescence correlation spectra contour maps showed obvious difference between different grades,and the predictive results based on the leave-one-out method showed the strong ability of this spectrum method for tea classification.