SummaryThis study proposes an improved energy‐based approach for quantitative classification of velocity‐pulse‐like ground motions. The pulse amplitude is determined, in its value and in time location, by the amplitude of the half‐cycle pulse having the largest seismic energy. After conducting statistical analyses, a newly‐determined threshold level for selecting pulse‐like ground motions is derived; and then what followed is a comparison analysis of three pulse‐detecting schemes, one using the wavelet analysis, the other two using the energy concept. It is believed that other than providing a useful way of classifying pulse‐like ground motions for structural demand analysis, knowledge of this work could also benefit the development of the ground motion prediction equations accounting for pulse effects, and further to aid the probabilistic seismic hazard analysis in a near‐fault environment. Copyright © 2016 John Wiley & Sons, Ltd.