This paper proposes a substructural damage identification approach without the information of responses and forces at the interface degrees-of-freedom. It is based on the response reconstruction technique using the unit impulse response function in the wavelet domain. The finite element model of the target substructure and acceleration measurement data from the damaged substructure are required in the identification. A dynamic response sensitivity-based method is used for the substructural finite element model updating, and local damage is identified as a change in the elemental stiffness factors. The adaptive Tikhonov regularization technique is adopted to improve the identification results with the measurement noise effect. Numerical studies on a three-dimensional box-section girder are conducted to validate the proposed method of substructural damage identification. The simulated damage can be identified effectively even with 10% noise in the measurements and a 5% coefficient of variation in the elastic modulus of material of the structure.