Abstract: Chemical industrial parks (CIPs) often exhibit groundwater plumes that are characterized by high concentrations, multicomponent mixtures, and pronounced temporal variability owing to intensive raw-material handling, complex production emissions, and accidental releases. These composite plumes pose cumulative risks to both ecosystem integrity and public health. Over the past two decades, groundwater-risk assessment has undergone three paradigm shifts-from static concentration-based thresholds, through multi-pollutant weighted scoring, to dynamic prediction frameworks that fuse process models with data-driven methods. This review systematically dissects the source-migration-exposure-effect chain of multi-contaminant groundwater in CIPs, summarizes recent advances in cross-scale simulation of coupled hydrological, geochemical, and pollutant migration-transformation as well as ecotoxicological processes, and highlights emerging risk-governance paradigms empowered by machine learning and digital-twin technology. Finally, future directions are proposed for regulatory implementation, data sharing, and interdisciplinary integration, providing a comprehensive reference for dynamic multi-pollutant risk assessment and precision management.