We present an implementation of the Polarizable Continuum Model (PCM) in combination with the Second–Order Polarization Propagator Approximation (SOPPA) electronic structure method. In analogy with the most common way of designing ground state calculations based on a Second–Order Møller-Plesset (MP2) wave function coupled to PCM, we introduce dynamical PCM solvent effects only in the Random Phase Approximation (RPA) part of the SOPPA response equations while the static solvent contribution is kept in both the RPA terms as well as in the higher order correlation matrix components of the SOPPA response equations. By dynamic terms, we refer to contributions that describe a change in environmental polarization which, in turn, reflects a change in the core molecular charge distribution upon an electronic excitation. This new combination of methods is termed PCM-SOPPA/RPA. We apply this newly defined method to the challenging cases of solvent effects on the lowest and intense electronic transitions in o-, m- and p-nitroaniline and o-, m- and p-nitrophenol and compare the performance of PCM-SOPPA/RPA with more conventional approaches. Compared to calculations based on time-dependent density functional theory employing a range-separated exchange-correlation functional, we find the PCM-SOPPA/RPA approach to be slightly superior with respect to systematicity. On the other hand, the absolute values of the predicted excitation energies are largely underestimated. This – however – is a well-know feature of the SOPPA model itself and is not connected to its combination with the PCM.