Bridges are one of the most vulnerable components of a highway transportation network system subjected to earthquake ground motions. Prediction of resilience and sustainability of bridge performance in a probabilistic manner provides valuable information for pre-event system upgrading and post-event functional recovery of the network. The current study integrates bridge seismic damageability information obtained through empirical, analytical and experimental procedures and quantifies threshold limits of bridge damage states consistent with the physical damage description given in HAZUS. Experimental data from a large-scale shaking table test are utilized for this purpose. This experiment was conducted at the University of Nevada, Reno, where a research team from the University of California, Irvine, participated. Observed experimental damage data are processed to identify and quantify bridge damage states in terms of rotational ductility at bridge column ends. In parallel, a mechanistic model for fragility curves is developed in such a way that the model can be calibrated against empirical fragility curves that have been constructed from damage data obtained during the 1994 Northridge earthquake. This calibration quantifies threshold values of bridge damage states and makes the analytical study consistent with damage data observed in past earthquakes. The mechanistic model is transportable and applicable to most types and sizes of bridges. Finally, calibrated damage state definitions are compared with that obtained using experimental findings. Comparison shows excellent consistency among results from analytical, empirical and experimental observations.