The article presents a new algorithm for resolving the fluorescence spectra of three-component systems. The proposed method refers to a technique already known in the literature combining self-modeling based on singular value decomposition with numerical optimization of the error function derived from the linear Stern–Volmer equation (SM-SVD-SV). Previously, this method was successfully used for resolving the fluorescence spectra of two-component systems. A formal extension of mathematical concepts employed by the SM-SVD-SV method to the case of three-component mixtures of fluorophores, even though possible, leads to expressions which because of its complexity are ineffective in the optimization process and thus have to be simplified. To this purpose, the suitably developed two-component difference fluorescence spectra are used to a broad extent. This way the search for the fluorescence spectra of pure components performed in the 3D space spanned by three significant eigenvectors is reduced to much easier and efficient searches in the 2D space. Further reduction of the number of the optimized parameters is achieved by a neat application of the rank annihilation factor analysis (RAFA) providing convenient estimates for the searched values of the Stern–Volmer constants of all three fluorophores. The effectiveness of the developed technique was tested on model spectra to which a homoscedastic noise was added. The real data set including the quenched fluorescence spectra of a three-component mixture of 9,10-dicyanoanthracene, 10-chloro-9-cyanoanthracene and 9-cyanoanthracene in methanol was also analyzed. The obtained results compared with those yielded by PARAFAC and direct hard modeling allow to consider the proposed algorithm as an useful tool that facilitates resolution and refinement of the spectra of pure components derived from suitably quenched mixtures of three fluorescing species.