The dual fluorescence of 4-(dimethylamino)benzonitrile (DMABN) has been intensively studied in the last decades, but surprisingly there is not any detailed theoretical study of its photochemistry in polar solvents. In this work, we rationalize the different luminescent behavior of 4-aminobenzonitrile (ABN) and DMABN in acetonitrile by a computational study developed at the CASSCF/CASPT2 level and using the Polarized Continuum Model to reproduce the solvent environment. We present here the critical geometries, energies, and connections between the potential energy surfaces of the low-lying excited states: the locally excited state (LE) and several intramolecular charge transfer states (ICT). The computational results show that the topology of the potential energy surfaces (PES) does not change substantially when the effect of a polar solvent is included, in comparison with the gas phase. For DMABN, though, polar solvents stabilize preferentially the ICT states in such a way that the different interplay with the LE state induces strong qualitative changes in the photochemistry of this compound. Specifically, the planar ICT (PICT) species located on the S2 surface in the gas phase is, in acetonitrile, located on the S1 surface. that is, at the geometry of the PICT minimum, the LE state is higher in energy than the ICT. Now LE and PICT minima are practically degenerate and, given that both correspond to first excited state species, emission can take place from both of them. However, the twisted ICT (TICT) species is still the most favored thermodynamically so it is expected that this species would be preferentially populated. On the other hand, for ABN the equilibrium lies in favor of LE, as the TICT species was found at a much higher energy with a low reaction barrier toward LE. This explains why dual fluorescence cannot be observed in ABN, even in polar solvents.