Abstract: Flooding is one of the main natural disasters that affect agricultural production. The loss assessment of crop flood damage is an important basis for guiding the prevention and mitigation of agricultural disaster, which is of great significance for maintaining the stability of agricultural production and ensuring food security. Firstly, we briefly reviewed the research advances on the formative mechanism in crop flood damage, sorted out the main methods of loss assessment of crop flood damage, and pointed out the advantages and limitations of these methods in various application scenarios. Secondly, based on the comments above, we proposed an integrated framework for dynamic assessment of crop flood damage. This framework is supported by multi-source data, powered by hydrodynamic model, based on the combination of numerical simulation and machine learning, and oriented towards disaster events. Finally, we proposed some practical suggestions and research directions for further study, including (1) constructing multi-parameter loss function and model libraries for different exposures and vulnerability environments, (2) strengthening the application of multi-source data to improve assessment accuracy, and developing parallel computing and artificial intelligence algorithms to improve computational efficiency, (3) constructing comprehensive platform for flood disaster assessment and changing to intelligent applications.