A novel control theory-based eigenstructure assignment (ESA) technique is employed to update the finite element model (FEM) of a linear time-invariant system. The proposed method uses state feedback to produce the gain matrix which in turn updates the existing system matrices through simultaneous assignment of eigenvalue-vector pairs in the FEM generated system matrices. However, unlike general control technique no external input is used to control the system. A subspace identification algorithm is applied to develop the state space model and the eigenstructure of its state matrix is used as the target for the ESA algorithm to update the FEM. Unlike most FEM updating algorithms which use optimization techniques, this method is not iterative (and hence computationally less expensive) as the state matrix gets updated just once yielding the single best possible solution to match the desired eigenstructure. Furthermore, it naturally preserves the basic exploitable properties like symmetry, sparsity, positive definiteness of the updated matrices. The method is first validated numerically on a four-story shear frame subjected to ambient vibrations. Following this, a two-story one-bay aluminium frame is subjected to a suite of excitations in the laboratory, and the response time histories are put through canonical variate analysis algorithm to yield the real system state space model, and hence the modal parameters. Noise is taken care of by singular value decomposition of the signal and performing ensemble averaging of the state matrix in its companion form. A very close conformation of the updated model is observed when compared to the modal parameters extracted from structural response.