Modern industrial development is accompanied by the increasingly frequent occurrence of accidental release atmospheric dispersion events, causing extremely serious human and property losses and environmental pollution, in which rapid and accurate prediction of atmospheric dispersion is an important task to mitigate the unexpected consequences. In this paper, we take the case of previous years as the starting point, firstly, the occurred hazardous chemical atmospheric dispersion accidents in the past five years are shown, and the related concepts of hazardous chemical atmospheric dispersion are given. Then, the current state of atmospheric dispersion research is reviewed, well-known experiments on atmospheric dispersion of hazardous chemicals are summarized, and correspondingly the existing atmospheric dispersion prediction models are classified into simplified-experience models, mechanism- and rule-driven models and data-driven models. In particular, for the purpose of rapid atmospheric dispersion prediction, some research on atmospheric detection and identification are analyzed in detail. Moreover, the relevant professional software for atmospheric dispersion prediction are introduced, and also their calculation adaptabilities regarding time-consumption and output accuracy are discussed. Thereinafter, according to the shortcomings of existing atmospheric dispersion prediction models in research and application fields, the development trend of atmospheric dispersion prediction research and technology is foreseen, and some feasible future research directions are proposed as follows: (1) the fusion of image processing techniques, the establishment of a database of historical accident scene information and meteorological information, (2) new correction algorithms, and (3) an emergency response system for full-scene atmospheric dispersion prediction.