Sizeable aftershocks following a major earthquake in short succession can lead to cumulative damage and an increase in permanent lateral deformations in a building, making it crucial to consider aftershocks in structural vulnerability assessments. Although numerous studies have considered the response of structures under mainshock-aftershock (MS-AS) sequences, there currently does not exist a standardized method for selecting or artificially generating aftershock records. Due to often limited data availability of large magnitude MS-AS records, sequences have commonly been generated by repeating mainshocks as aftershocks, or selecting records that can represent an aftershock based on empirical laws. Such relationships include the Gutenberg-Richter magnitude-recurrence law, Bath’s law for the MS-AS magnitude relation or Omori’s law for aftershock frequency. This paper aims to assess how three aftershock selection criteria commonly utilized in literature (replicate, random, and generalized Omori’s law) influence seismic fragility by comparing their results to natural MS-AS sequences. A model three-story moment-resisting steel frame building is developed and subjected to MS-AS loading based on the different record selection methods. The results showcase how poorly selected aftershock ground motions can misrepresent the seismic demands.