In this paper, a two-stage damage identification method is proposed using the data obtained from limited piezoelectric sensors. In the first stage, a first-order approximate technique is proposed for numerically obtaining the transient response of the change in electrical potential on sensors caused by damages. To identify damages, both numerical and experimental data in the time domain are transformed into the frequency domain using the FFT technique. The damage locations, then, can be detected by matching the numerical data and the experimental data in the frequency domain through a proposed detection technique, which can eliminate the effect of damage extents effectively and keep the influence of damage location only. After obtaining the possible damage locations, in the second stage, an iterative estimation scheme for solving nonlinear optimization programming problems, based on the quadratic programming technique, is proposed to predict damage extents. A beam example is employed to illustrate the effectiveness of the present algorithm. Furthermore, various investigations, such as the accuracy of the proposed first- order approximate technique, the influences of the excitation frequency of external forces, modeling errors and measurement noises and window methods used in the FFT have been carried out.