In this paper Group Method of Data Handling (GMDH) technique is implemented and applied to the system identification problem of a Magneto-Rheological (MR) damper. GMDH networks are used to approximate the forward and inverse dynamic behaviors of a MR damper. The difficulty of this identification problem stems from the inherent hysteretic and highly nonlinear dynamics of MR dampers. The GMDH-based solution exhibits comparable identification accuracy in both the forward and inverse models as compared to state-of-the-art more computationally extensive solutions. © 2009 IEEE.