Abstract: Flash floods constitute the most fatal natural disasters in China, with their prevention representing the most critical deficiency in current flood risk management frameworks. Accurate simulation and scientific early warning are key to effectively reducing flash flood disaster losses. This paper systematically reviews the research progress in flash flood simulation and warning models, analyzing the current application status and development prospects of physical mechanism models such as hydrological models and hydrodynamic-sediment models, data-driven models such as machine learning, as well as warning modes including meteorological risk warning, critical rainfall warning, and water level (discharge) warning. In response to key challenges faced in flash flood simulation and warning research, including physics-data fusion, construction of refined warning systems, compound disaster chain warning, and global flash flood mega-models, this paper proposes future development directions for flash flood simulation and warning technologies.