To date, mathematical models for wave energy devices typically follow Cummins equation, with hydrodynamic parameters determined using boundary element methods. The resulting models are, for the vast majority of cases, linear, which has advantages for ease of computation and a basis for control design to maximise energy capture. While these linear models have attractive properties, the assumptions under which linearity is valid are restrictive. In particular, the assumption of small movements about an equilibrium point, so that higher order terms are not significant, needs some scrutiny. While this assumption is reasonable in many applications, in wave energy the main objective is to exaggerate the movement of the device through resonance, so that energy capture can be maximised. This paper examines the value of adding specific nonlinear terms to hydrodynamic models for wave energy devices, to improve the validity of such models across the full operational spectrum.