Hydrodynamic dynamometers are employed in internal combustion engine (ICE) test benches for the entire range of ICEs, from cart engines to large ship engines. They are inexpensive and characterized by a relatively small moment of inertia. Although nonlinearities and the absence of accurate models make controlling these dynamometers a difficult task, they are currently used almost exclusively for stationary measurements. To overcome this limitation, this paper proposes an inverse torque control based on a Wiener approximation whose parameterization is strongly simplified by choosing a suitable set of basis functions for the nonlinear map. As inverting this nonlinear map would be computationally demanding, it is replaced with a dynamic inversion which-together with the actuator redundancy-allows additional optimality criteria to be incorporated. The performance of the proposed control law is validated by measurements on a test bench and shows significant improvement compared with classical implementations. © 2013 IEEE.