The most suitable wavelength intervals were selected for the determination of 4 polycyclic aromatic hydrocarbons (PAHs; benzo[g,h,i]perylene, dibenzo[a,h]anthracene, pyrene, and triphenylene) in very complex mixtures of 11 PAHs: anthracene, benz[a]anthracene, benzo[a]pyrene, benzo[b]fluoranthene, benzo[g,h,i]perylene, benzo[k]fluoranthene, chrysene, dibenz[a,h]anthracene, phenanthrene, pyrene, and triphenylene. The multiple linear regression algorithm was applied to measurements made in several wavelength intervals previously selected on the basis of sensitivity and minimum number of interfering compounds. Of the different models obtained, those displaying minimum error propagation in the analytical result were selected. By applying the models proposed in this study, we precisely and accurately determined benzo[g,h,i]perylene, dibenz[a,h]anthracene, pyrene, and triphenylene in complex mixtures—a feat that could not be achieved by the use of constant-wavelength spectrofluorimetry in combination with second-derivative techniques.