Curved bridges are constructed to conform to geometric constraints resulting from traffic and structural restrictions. They are different from their straight counterparts since the response coupling in the longitudinal and transverse directions and rotation of the superstructure may lead to significantly different seismic response. Observations from past earthquakes highlighted the seismic vulnerability of these bridges due to this coupled response. The consequence of bridge damage on the performance of transportation system is commonly assessed through Seismic Risk Assessment (SRA) of lifeline systems. Thus, seismic fragility curves are essential input to SRA to estimate damage to highway bridges and consequently to the network. The literature review shows shortcomings in fragility studies on the effect of horizontal curvature of bridges, specifically concrete box-girder bridges. This study aims to fill in the gap on the current state-of-the-knowledge in the seismic response and vulnerability of curved concrete box-girder bridges. Since this bridge type is common in California, the modern details adopted by CALTRANS along with the current seismic design considerations from SDC (2013) are used to select the representative benchmark bridges. To incorporate the uncertainty in geometrical, structural, and material properties of bridges into the analytical models, five sets of statistical bridge samples (each includes 160 bridges) with various subtended angles are developed. These bridge models are subjected to four sets of ground motions representing different site soil conditions and spectral characteristics. A total of 800 response history analyses are performed and the results are used to develop analytical component and system fragility functions for a range of subtended angles. A comprehensive study on the effect of horizontal curvature on the bridge dynamic characteristics and component seismic response is conducted. The median of system (bridge) fragility curves are proposed as a function of the subtended angle for each ground motion set. These functions can be used as input into SRA tools. The fragility analysis shows that the seismic vulnerability of bridges depends on the soil condition of the site and ground motion characteristics as well as the horizontal curvature of the bridge. Columns are found to have the most significant contribution to the system fragility curves. The analyses confirm that the current seismic details including PTFE/spherical bearings and isolated shear keys, suggested by CALTRANS, achieve the objectives of capacity-protected design of piles. Since the dynamic characteristics of bridges are sensitive to the curvature, curved bridges with subtended angles greater than 30 degrees require explicit modeling of curved geometry. In curved bridges, the coupling of transverse and longitudinal modes reduces the dominance of the fundamental mode in the bridge response and leads to the contribution of higher modes. The statistical evaluation of structural demands indicates that the curvature and the torsion demands on columns are amplified in curved bridges.