The development and spread of fire incidents for oil storage system will lead to more serious secondary incidents. In order to grasp the possible development process and changing trend of the incident more accurately, based on the scenario analysis approach, this paper proposes a scenario evolution method of oil storage system fire based on historical cases and GERT (Graphical Evaluation and Review Technique) network, and develops a scenario evolution GERT network model. Through the statistical analysis of 173 historical cases of fire and explosion incidents in oil storage tank at home and abroad in nearly 50 years, twenty main scenarios in the emergency processes for fire incidents in oil storage system are summarized, and the possible scenario evolution paths of fire incident for oil storage system are developed and transformed into GERM network of scenario evolution. The probability of occurrence, average occurrence time, variance of occurrence time and other characteristics of possible paths of scenario evolution are obtained by analysis algorithm of GERT network. Among them, the solving rule of scenario occurrence probability is put forward, which combines the scenario conditional probability obtained by case statistics with the scenario experience probability obtained by expert score modified by DS (Dempster-Shafer) evidence theory. Taking the“7·16”Dalian Oil Depot fire and explosion incident as an example, the effectiveness of the proposed method and model is verified by comparing the development and evolution process with actual event. The scenario evolution method and model of oil storage system fire based on historical cases and GERT network can better solve the problem of qualitative and quantitative description of the development and evolution process of the incident, which provides a new quantitative analysis method for judging the scenario evolution of fire incident in oil storage system from the perspective of probability statistics.