Abstract Currently, there is no research effort on the efficient selection of earthquake records for the non-uniform excitation of bridges. The present investigation addresses this topic. Initially, the physics of the bridge response to asynchronous seismic excitations are illustrated. Next, two basic criteria are presented to provide insight into the selection of reference ground motions for the simulation of asynchronous excitations. The first criterion is the ratio of the zero-crossings of the reference ground displacement time series (n) to the total duration of the ground shaking (t), which can be evaluated by visual inspection of the displacement time series, and the second is the peak ground displacement. The former criterion is an indicator of the degree of sensitivity of the simulated displacement time series at various separation distances to the loss of coherency, and the numerical results of the present study indicate that the simulated waveforms of earthquake displacements with n/t greater than unity are highly dependent on coherency, i.e. become significantly modified by coherency, and, consequently, are more crucial for the pseudo-static response of bridges subjected to asynchronous seismic excitations. The latter criterion is also directly related to the magnitude of the pseudo-static force induced in the bridge structure. Finally, a brief discussion on the selection strategy of earthquake records for the seismic assessment of bridges is provided.