In this paper, a damage detection method based on modal parameters, which deals with incomplete measurement, is presented. Intact and damaged structures are modeled using the finite-element method. Damage in structure is assumed as a reduction in elastic modulus. This study utilizes an approximate stiffness matrix of damaged structure to improve the data expansion methods, which is used to complete the incomplete modal parameters of damaged structures. Verification of proposed method is done using a 2D truss and a 2D frame. This method is able to expand the measured data and to solve the sensitivity equation, which depends on complete mode shapes and frequencies of intact and damaged structure, and to detect the damage in the presence of noises, such as measurement and modeling errors. Results show that damage detection using improved data expansion is more accurate and reliable than damage detection using common data expansion.