Over the past decade, seismically induced damage to bridges has been widely reported following major earthquakes such as the 1994 Northridge, 1995 Kobe and 1999 Chi-Chi events. Since these earthquakes, restrainers and stoppers have been installed on bridges to prevent unseating and excessive displacements, respectively. Alternatively, column jacketing has also been proven to be effective. However, the enhanced shear strength may result in extra retrofitting works on the footing. For bridges damaged in the Chi-Chi earthquake, investigations revealed that most bridge columns experienced none-to-minor damage in the longitudinal direction. The reason for this unexpected performance was the construction practice of using a rubber bearing, which is an unbolted design that may slide under large lateral forces. In this paper, parametric studies on simply-supported bridges retrofitted by a restrainer or concrete shear key along the longitudinal and transverse axes were carried out. The research focuses on finding suitable combinations of the design force and gap spacing so the restrainer and concrete shear key can be used as an unseating prevention device, with respect to the allowable column damage in terms of displacement ductility under near-facult type earthquakes. A two-lane PCI-girder bridge was selected as the benchmark model. In the longitudinal direction, a total of nine combinations considering yielding strength and gap spacing for the restrainer were analyzed; while parameters for the concrete shear key were divided into three shear force levels and three gap spacings. In the transverse direction, a similar approach was adapted, except smaller gap spacing was used. For each of the above mentioned earthquakes, seven input ground motions were selected and their PGAs were adjusted to 0.36g and 0.45g as the Design earthquake and Maximum Considerable Earthquake, respectively. Based on the results of nonlinear time history analyses, proper parameters to design the restrainers and concrete shear keys are obtained. Responses obtained from numerical simulations under the Chi-Chi earthquake leaded to new implications to design those devices. Restrainer should not exceed its breaking strain and sufficient unseating length will be needed always. Concrete Shear key was determined by considering both displacement demand of the superstructure and displacement ductility of the column at the same time. Further study is needed to provide optimal design parameters for use in performance based bridge design.