: In recent years, ground motion models used in probabilistic seismic hazard analyses (PSHA) have evolved from the traditional approach of using ground motion prediction equations (GMPEs) to using ground motion time series models. The purpose of this paper is to develop an approach to perform a probabilistic seismic hazard analysis using stochastic site-based simulation techniques. These techniques consist of empirical stochastic models that simulate both near-fault and far-field ground motion time series. The near-fault models consider directivity pulses, which can impose large seismic demands. The proposed approach was applied to a site located in Los Angeles Downtown and the corresponding hazard curves were developed. The results were compared to hazard curves derived for the same site from CyberShake, which uses a physics-based simulation approach, and from a traditional GMPE approach. The comparison indicated that the proposed methodology accurately describes the seismic hazard at the site at high hazard levels. The proposed approach is computationally efficient compared to the use of physics-based simulations like CyberShake.