Magnetostrictive materials, such as Terfenol-D, are used in manufacturing high efficiency actuators. Precise positioning control for such actuators can be achieved by simply controlling the input current. However, modeling of such actuators using physical principals often leads to a set of complex nonlinear equations, not suitable for control system design. Therefore, there is a need to use simple empirical models, along with parameter identification algorithms, appropriate for designing low complexity controllers. In this paper, an actuator system consisting of Terfenol-D (as active element), a magnification mechanism, and a Peltier thermoelectric cooler (TEC) is used as a case study. Input/output data from the physical system is generated using Hardware-in-the-Loop (HIL) technique. This data is then used to develop linear model that captures the dynamical behavior of the actuator. The Auto- Regressive Moving Average (ARMA) model was selected. Its parameters were identified using the recursive least squares algorithm. The optimal model, in terms of accuracy and complexity, is then selected for experimental validation. Initial simulation and experimental results show that linear models can be good candidates to be used as a basis for position control system design.