In this paper, we present undamaged agent based, an effective damage identification technique along with Eigen-system Realization Algorithm (ERA) for damage assessment in laminated composite structures. Unlike the use of frequency measurements, this technique has the potential to readily pinpoint the location and the extent of damage, in addition to alerting to its occurrence. The proposed technique is implemented through numerical simulation and experimentation on laminated beams with and without delaminations. For numerical simulation, a Sate-space based finite element system model (SS-FESM) is employed to generate transient impulse response signals. In this study, we employ ERA for system identification of fibre reinforced composite laminates to identify the presence of delaminations and assess their location and severity. The impulse response and the input signals are feed into ERA to estimate the normalized dynamic properties, namely the stiffness and damping coefficients of all elements in the structure. Variations in the dynamic properties readily identify the presence of damage, however, the location and severity of the damage is assessed or determined through undamaged agent based iterating technique. The numerical simulation of 8 ply quasi-isotropic Carbon Fiber Reinforced Polymer (CFRP) laminated beams demonstrate the delaminations as small as one percent of the beam length in size can be identified and assessed using the present method. Experimental simulations indicate that noise in the measured impulse signals need to be eliminated for successful practical implementation of the technique.