Fluorescence technology is an effective tool for detecting polycyclic aromatic hydrocarbons (PAHs) in water. However, the accuracy of fluorescence detection is reduced by the spectral overlap of different PAHs and coexisting colored dissolved organic matter (CDOM). In this study, a single-excitation interval selection method based on an excitation-emission matrix is proposed to quantify four PAHs: fluorene, pyrene, phenanthrene, and benzo(a)pyrene under CDOM interference. The optimal excitation wavelength for each PAH was obtained by stability analysis, based on which the optimal emission interval was obtained by chaotic particle swarm optimization. The partial least squares (PLS) prediction models of four PAHs under interference were established. On comparing with other modeling methods, the results show that the models with interval selection have better prediction accuracy (mean relative error < 10%) under CDOM interference. The recovery rate and limit of detection of the method were also evaluated. This study provides a new and helpful strategy for fluorescence detection of interfering PAHs in water.