The analysis of multi-component three-dimensional fluorescence overlapping spectra is always very difficult. In view of the advantage of differential spectra and based on the calculation principle of two-dimensional differential spectra, the three-dimensional fluorescence spectra with both excitation and emission spectra is fully utilized. Firstly, the excitation differential spectra and emission differential spectra are respectively computed after unfolding the three-dimensional fluorescence spectra. Then the excitation differential spectra and emission differential spectra of the single component are obtained by analyzing the multi-component differential spectra using independent component analysis. In this process, the use of cubic spline increases the data points of excitation spectra, and the roughness penalty smoothing reduces the noise of emission spectra which is beneficial for the computation of differential spectra. The similarity indices between the standard spectra and recovered spectra show that independent component analysis based on differential spectra is more suitable for the component recognition of three-dimensional fluorescence overlapping spectra.