Abstract: Flash flood forecasting is a worldwide hot issue of flood control and disaster mitigation. Based on an analysis of previous research, this paper reviews its technical connotation and key aspects, including data acquisition, model calculation, index determination, and uncertainty analysis for the early warning of flash flood. In the previous studies of flash flood forecasting, we can find several shortcomings: the comprehensive analysis of multi-source warning data needs to be strengthened; the research on the rainfall-runoff relationship under complicated underlying surfaces is not deep enough; the accuracy of flash flood forecasts is still poor; how to improve the predictability of such forecasting needs further research. We suggest three directions for future studies: tackling the mechanism of flash flood formation, constructing a stereoscopic perception forecasting system of flash floods, and determining an index threshold of early flood warning.