Abstract: Multi-point ground motion excitation input is essential for the seismic analysis of large-span spatial structures, where the key challenge is the accurate matching of target seismic spatial variability characteristics in the simulation of multi-point ground motion time histories. To address this issue, this paper systematically reviews the development of multi-point ground motion synthesis methods and their engineering applications, with a focus on the practical demands of seismic analysis for large-span spatial structures. First, the fundamental differences between unconditional and conditional simulation methods are discussed in terms of theoretical assumptions, implementation strategies, and applicable scenarios, highlighting the evolutionary trend of multi-point ground motion synthesis from purely statistical or empirical models toward approaches incorporating observational constraints and physical mechanisms. Subsequently, based on studies of representative large-span spatial structures such as bridges, tunnels, and transmission towers, the influence of seismic spatial variability on structural dynamic response analysis, system-level risk assessment, and resilience evaluation is summarized. The seismic spatial variability can significantly amplify structural responses and alter failure modes, while traditional uniform ground motion excitation input may underestimate the actual structural responses and actual seismic risk of structures. This research lays a foundation for developing performance-based design theories and risk assessment frameworks for large-span spatial structures with spatial variability considered.