Objectives  In recent years, atmospheric dispersion models have become important research contents in environmental assessment, disaster prevention and mitigation, and emergency management. At present, mainstream environmental quality regulations models include the AERMOD model and CALPUFF model recommended by the U.S. Environmental Protection Agency (EPA). These models are widely used in environmental assessment and atmospheric environmental quality assessment, but rarely used in toxic gas dispersion simulation and decision support in emergency response. The objective of this article is to compare the differences between the two models and give ideas for future development of expanding the atmospheric dispersion model to the field of emergency response.

  Methods   First, the basic theories of the two models are introduced and analyzed. Then, four comparative experiments are designed and carried out. Experiment 1 is designed to observe whether the pollutants existing in the original wind direction will accumulate to the next moment and affect the new concentration distribution after the sudden change of wind direction in the two models. Experiment 2 is designed to observe the processing capacity of the two models for static wind. Experiment 3 is designed to observe the results of the two models in two different dispersion scales. Experiment 4 is designed to compare the calculation speed of the two models after greatly increasing the spatial resolution.

  Results   Experiment 1 shows that the simulation results of the CALPUFF model for sudden changes in wind direction are more reasonable than the AERMOD model. Experiment 2 shows that the simulation results of CALPUFF for special wind fields, such as breeze, are more reasonable than AERMOD. Experiment 3 shows that AERMOD is not suitable for cases where the experimental range exceeds 50 km, and within 50 km, the simulation results of both models are acceptable. Experiment 4 shows that after greatly improving the mesh resolution, the time‐consuming increase in AERMOD is not large, while the time‐consuming of CALPUFF increases significantly. Considering the calculation accuracy, three‐dimensional terrain and meteorological parameters, the CALPUFF Gaussian puff model has more advantages. At present, the integration of GIS and CALPUFF is more common and the application range is wider, but its disadvantage is that the calculation speed is slower than AERMOD.

  Conclusions   When the demand for calculation speed is low, the AERMOD model should be given priority. And when the requirements for spatial‐temporal resolution and calculation accuracy are relatively high, the CALPUFF model should be given priority. Under normal circumstances, emergency response simulation is preferred to use CALPUFF. Existing atmospheric dispersion models perform well in air pollution prediction and environmental assessment. However, the research of the models in the application of emergency management is not deep enough. Future improvements include higher calculation accuracy, higher running efficiency, more theoretical support for complex application scenarios, and the integration of multiple models to form a dedicated model library and model service chain for emergency management.