This paper presents stuctural damage detection method based on changes in Transfer Function Estimate (TFE) for detecting damage, predicting its location and monitoring the growth in damage. This method assumes that the displacement or the acceleration response time histories at various locationsalong the structure both before and after damage are available for damage assessment These responses are used to estimate TFE. The change of TFE between the baseline state and the current state is then used to identify the location of possible damage in the structure. The method is applied to the experimental and numerical data obtained from a railway steel bridge that is no longer in service. Several damage scenarios were introduced to the main girder of the test structure. Results show the method can be used to detect the damage, determine the exact location and measure the growth in damage with very good accuracy. The use of piezoelectric actuators as a local excitation source for large structures such as steel bridges is also presented Experimental and numerical results show that the Proposed approach may be successfully implemented on-line to detect the damage and to locate regions where damage occurred