Abstract: The Eigensystem Realization Algorithm (ERA), as a model parameter identification method in time-domain has received more and more attention in recent years. In this paper, the characteristics of the algorithm for identification of displacement and acceleration data in free vibration at the presence of the white noise is studied and compared. It is found that the displacement data and the acceleration data are sensitive to the lower and higher modes respectively when the standard ERA is employed for identification. The tracing capability of the algorithm to the vibration level of the system is also studied with the aid of the sliding window technique, and a method for identifying aliasing frequency is established. In addition, the ERA is extended to the parameter identification of a system with constantly forced response data. The corresponding formulations are presented and verified with several examples.