Considerable effort has been expended on the design of control systems for wave energy converters (WECs) over the past two decades. Working from the fundamental requirement of impedance matching, a variety of conceptual and practical algorithms have emerged, which bring various levels of realism to the original complex conjugate ideal, facilitating maximum power transfer. Simplifications can be introduced, such as passive control and causal control, while some enhanced algorithms allow physical constraints to be observed, nonlinear model dynamics to be articulated, or nonideal power takeoff systems to be recognized. However, in general, model-based WEC control systems are evaluated in tandem with identical simulation models, while the sensitivity of controller performance to modeling errors is ignored. In addition, the WEC model utilized by the controller rarely, if ever, fully represents the nonlinear nature of the true WEC dynamics. This paper articulates this model sensitivity issue for different WEC control system architectures and shows that it is of potentially greater impact than for traditional regulation/servomechanism control problems. Recommendations are given on the best WEC control architecture to adopt from a sensitivity/robustness perspective.