In this paper, a new approach for damage detection using a chaotic signal as an input excitation and steady state attractor-based measures as diagnostic parameters is investigated by means of numerical simulations. The method utilizes the deterministic, extreme sensitive properties of the chaotic signal to give rise to a low-dimensional response for feature extraction. This approach is applied to two numerical examples, the 4 DOF spring-mass-damper and a cantilevered beam system, where the damage is produced by varying the structural damping and stiffness, respectively. Lyapunov dimension is calculated as a “feature” for detecting the damage. Results show that this approach is feasible to detecting structural damage.