In this study, various aspects of vibration-based damage assessment of bridge structures are investigated. Results of application of novel methodologies in vibration-based damage assessment of structures in the presence of wide range of real seismic damage and from noisy and incomplete measurements are presented and discussed. The first part of this dissertation is devoted to experimental modal analysis. Two output-only and input-output system identification techniques are applied for identification of modal properties of the bridge from ambient vibrations and responses to high amplitude earthquake excitations respectively. An optimization-based finite element (FE) model updating methodology is applied for identification of damage characteristics by monitoring the variations in stiffness properties of critical elements of the bridge. A hybrid optimization procedure based on Genetic Algorithm (GA) and quasi-Newton optimization techniques is implemented for finding the best set of FE model parameters that minimizes the objective functions. Two objective functions are defined expressing the discrepancy between the measured and analytical response characteristics in time and modal domains. The meaningful agreement between FE model parameters identified using time and modal domains data with experimental stiffness indices indicates the efficiency and accuracy of the proposed damage identification procedure. In the final part of the dissertation, two vibration-based procedures are presented and applied for investigation of the consequences of damage in collapse capacity and functionality status of the bridge. The first procedure takes advantage of a double-integration and filtering routine to estimate the maximum drift ratios experienced by the lateral force resisting elements of the bridge from acceleration measurements. Estimated drift ratios along with pushover curves of the corresponding elements are used to calculate the ductility-based residual capacity of the elements and the bridge. The second procedure utilizes the incremental dynamic analysis (IDA) curves for estimation of collapse capacity of the bridge. A new approach for generation of FE model realizations of the earthquake damaged structures is proposed and applied. Generated FE realizations of the damaged bridge are used to estimate the collapse capacity of the structure. Amount of loss in collapse capacity along with seismic hazard characteristics at the bridge site and a set of tagging criteria are utilized for tagging and determination of the functionality status of the damaged bridge. Accuracy and reliability of the residual capacity estimation procedures are evaluated and verified by the results of a shake table experiment on a large-scale model of a reinforced concrete bridge.