This paper introduces a 3-step approach for the identification of a linear structure that is controlled by nonlinear damping devices. First, the structure with the integrated nonlinear damper is subjected to random vibration test and the frequency response function (FRF) of the structure is calculated from the input-output data of the physical system. Based on the frequency domain data, a state space model is then estimated using a recently developed FRF curve-fitting technique that is designed especially for lightly damped structures with control inputs. Finally, an iterative process is used to optimize the model performance in the time domain and an integrated model of the nonlinenrly controlled structure is derived by interconnecting the structure model with that of the nonlinear damper. The complete approach is illustrated by the modeling of a base-isolated structure controlled by a magnetorheological (MR) fluid damper.