This dissertation systematically addresses the modeling, quantifying, and protection of highway bridges against earthquake hazards. Firstly, the research substantially improves the p-y spring based simulation method to predict the seismic responses of highway bridges that accounts for various soil-structure interaction effects. Closed-form formulae are provided for the p-y spring input parameters to capture the bridge-embankment interaction effects, based on which an integrated step-by-step modeling procedure is developed. The procedure is applied to simulate the seismic responses of a well instrumented highway overcrossing and validated against the recorded responses during the 1992 Petrolia earthquake. Secondly, the study derives a response modification factor to quantify the relative impact of liquefaction induced lateral spreading with respect to seismic shaking on column drifts for highway bridges. The column drift response under lateral spreading is correlated to the crust layer energy imposed on the pile foundation at bridge piers. Under seismic shaking, the column drift ratio is directly related to the peak ground acceleration. By normalizing the column drift under the lateral spreading to that of under the seismic shaking, the proposed modification factor captures key features of how columns respond under both lateral spreading and seismic shaking, and offers reliable column drift demand predictions. Thirdly, this study investigates the effectiveness and optimal design of seismic protective devices for highway bridges. Component-level fragility functions are developed by using the probabilistic seismic demand analysis. To transparently quantify the bridge performance at the system level, seismic repair cost ratios are derived to combine damage probabilities, damage ratios and replacement costs of critical bridge components. Thereafter, a multi-objective genetic optimization method with the Pareto optimal concept is employed to identify the optimal design parameters of protective devices. Subsequently, the research derives a consistent performance index to facilitate the performance-based design and optimization of seismic protective devices. By converting the system-level repair cost ratio to be a function of median-level engineering demand parameters, a uniform design surface is generated for various protection designs. The derived surface can be easily implemented in the performance-based seismic protection design and optimization without iteratively updating the design goal when a new group of design parameters are considered. The robustness of the proposed method is examined in a case study to identify the optimal protection designs by using a genetic optimization scheme. Lastly, the study derives the seismic demand models for bridge rocking columns with foundation on rigid supports when subject to horizontal near-fault strong motions. The system equations of motion are derived and solved to incorporate the column flexibility and the rocking impact mechanism. By representing the near-fault ground motions with corresponding pulses, dimensional analyses are carried out to regress the closed-form expressions of system's drift and uplift demands. A rigorous validation process is implemented to demonstrate that the proposed models can be used with confidence to predict the seismic demands of the rocking system directly from structural and ground motion characteristics.