The traditional ground-motion models （GMMs） do not account for pulse effects and may therefore fail to estimate seismic hazards and risk at near-fault sites, where pulse-like ground motions are expected. Thus, the applicability of the NGA-West2 GMMs to the near-fault velocity pulse-like ground motions need to be tested. The near-fault strong ground motions are quantitatively identified from recent earthquakes since 2013 by using wavelet method and taking the uncertainty of pulse orientation into consideration so as to form a new pulse database. Based on the new pulse database, long-period pulses are extracted from the original pulse records by using wavelet method. Based on a quantitative analysis of the epsilon parameter, we quantitatively test the applicability of the NGA-West2 ground-motion model to the near-fault velocity pulse-like ground motions. The results show that the four NGA-West2 models are more suitable for describing the residual recordings in the studied period, but underestimate the original pulse-like ground motions especially around the pulse period. We noted that, among the four NGA-West2 models, the applicability of the CB2018 to the residual ground motions is the best. This study provides an excellent opportunity to quantitatively evaluate the NGA-West2 GMMs and to update these models in the near future, and also provides a basis for incorporating pulse effects into near-fault probabilistic seismic hazard analysis and seismic design.