The mechanism of photoinduced electron transfer (PET) from the aromatic amino acids (Trp32, Tyr35 and Trp106) to the excited flavin mononucleotide (FMN) in the wild type (WT) and four single amino acid substitution isomers (E13T, E13Q, W32A and W32Y) of FMN binding protein (FBP) from the Desulfovibrio vulgaris (Miyazaki F) were simultaneously analyzed (Method A) with the Marcus–Hush (MH) theory and Kakitani–Mataga (KM) theory using ultrafast fluorescence dynamics of these proteins. In addition, the PET mechanism of the WT, E13T and E13Q FBP systems (Method B) were also analyzed with both MH and KM theories. The KM theory could describe all of the experimental fluorescence decays better than the MH theory by both Methods A and B. The PET rates were found to largely depend on the electrostatic energies between photo-products, isoalloxazine (Iso) anion and the PET donor cations, and the other ionic groups, and hence on static dielectric constants. The dielectric constant (ε^DA₀) around the PET donors and acceptor was separately determined from those (ε^j₀, j = WT, E13T, E13Q, W32Y and W32A) in the domain between the Iso anion or the donor cations and the other ionic groups in the proteins. The values of ε^DA₀ were always lower than those of ε^j₀, which is reasonable because no amino acid exists between the PET donors and acceptor in all systems. The values of the dielectric constants ε^j₀ (j = WT, E13T and E13Q) were similar to those obtained previously from the analysis of the crystal structures and the average lifetimes of these FBP proteins. Energy gap law in the FBP systems was examined. An excellent parabolic function of the logarithms of the PET rates was obtained against the total free energy gap. The PET in these FBP isomers mostly took place in the so-called normal region, and partly in the inverted region.