Polarizable continuum solvation models are nowadays the most popular approach to describe solvent effects in the context of quantum mechanical calculations. Unexpectedly, despite their widespread use in all branches of quantum chemistry and beyond, important aspects of both their theoretical formulation and numerical implementation are still not completely understood. In particular, in this perspective we focus on the numerical issues of their implementation when applied to large systems and on the theoretical framework needed to treat time dependent problems and excited states or to deal with electronic correlation. Possible extensions beyond a purely electrostatic model and generalizations to environments beyond common solvents are also critically presented and discussed. Finally, some possible new theoretical approaches and numerical strategies are suggested to overcome the obstacles which still prevent a full exploitation of these models.