The rainfall threshold is an important indicator of flash flood conditions. In this study, the existing methods for computing rainfall thresholds are divided into two categories and reviewed on the basis of their technical principles. The two categories include the data-driven statistical and inductive methods and the physical process-based hydrologic hydraulic methods. As expansions of rainfall thresholds, the dynamic rainfall threshold and the storm critical curve are also introduced and discussed together with advances in uncertainty analysis of rainfall thresholds. In our review, the statistical and inductive methods have been more widely accepted in China. Moreover, antecedent rainfall (or antecedent soil saturation) and cumulative rainfall at particular time intervals are the two governing factors commonly considered in the calculation of rainfall thresholds. Cumulative rainfall may be the loneliness factor to be considered at times. Further, it is found that the rainfall threshold conveys poorly the magnitude of flash flooding. Understanding of the uncertainty in rainfall threshold calculations would be helpful for the improvement of flash flood warnings. However, how to incorporate the uncertainty into the decision-making process still remains a major challenge.