AbstractA mechanical bar splice, which is known as a bar coupler, is an alternative to traditional lap splicing to connect bars in reinforced-concrete (RC) structures. Although it is feasible to incorporate mechanical bar splices in bridge bents as a precast detailing to expedite construction, the use of bar couplers in the plastic hinge region of bridge columns is prohibited in current US codes. This ban was mainly because the coupler performance and the effects of couplers on the seismic behavior of columns were not fully understood. Further, the test data regarding the performance of mechanically spliced bridge columns are scarce, and the available data are for columns with different geometries, confinements, and testing procedures. An experimental investigation was performed to systematically determine the seismic performance of mechanically spliced bridge columns and to develop the most comprehensive test database for these columns. Seven half-scale bridge columns were constructed and tested. One column was cast-in-place (CIP) to serve as the reference model and six were precast and incorporating different couplers at the column base with a coupler product per specimen. Two grouted, one threaded, and three hybrid couplers were included. To minimize the effects of other test parameters, all columns had the same geometry, were longitudinally reinforced with steel bars from a single batch, and were tested under the same loading. The test results showed that seismic couplers, which had a consistent failure mode of bar fracture, reduced the precast column displacement capacities from 3% to 45% compared with CIP and the lateral strength of the precast columns was slightly higher than that of CIP. All precast columns that were tested in this study met the current code seismic requirements; therefore, they might be used in any seismic region of the United States. However, only seismic couplers must be utilized in bridge bents. Finally, a simple design method for mechanically spliced bridge columns was evaluated using a database that consisted of 10 specimens and was viable for practice.