We present a theoretical modeling of the energy-loss spectroscopy data for monolayer graphene (MLG) supported by Pt(111), Ru(0001) and Ni(111) substrates. To reproduce the experimental loss function, we have used a two-dimensional, two-fluid hydrodynamic model for interband transitions of graphene's π and σ electrons and an empirical Drude–Lorentz model in the local approximation for metal substrates. The electronic response from the visible to the ultraviolet frequency range has been nicely reproduced for MLG/Pt(111) and MLG/Ru(0001). For graphene nanodomes on Ru(0001), the loss function of valleys shows an additional mode at 8–9 eV. By contrast, this models fails for the case of MLG/Ni(111), presumably due to the strong hybridization between the π states of graphene and the d bands of Ni, which is not accounted for in the model.