Abstract: The interaction between disasters,derived disasters and secondary disasters may make the disas ter situation worse and hard to control.Rush emergency,risk avoiding and controlling for the process of disaster rescue,as countermeasures to the disaster,should play their role in turning the damage caused by the disaster as low as possible.The paper focuses on describing the dynamic evolutionary process,before establishing a GERT(Graph Evaluation and Review Technique) network model,taking the natural evolu tion of disaster and the action of emergency rescue into consideration,based on the Bayes reasoning.After that,by means of combining of GERT network methods and Bayes reasoning tools,according to the receiv,d new information,the paper makes dynamic modification to active parameters in GERT network,and measures like dynamic prediction,earlyvarning and evaluation have been taken against different probalili ties and time of revolution in different situation.In the end,the paper gives qualitative and quantitative a nalysis to the influence to systematic revolution caus,d by external actions,like deriv,d disaster and see ondary disasters.In award,this paper provides the framework and tools for the disaster evolution in the way of qualitative and quantitative analysis,which reveals the mechanism of disaster evolution,at the same time,it offers a new method and idea for the forecasting,early warning and evaluation of the disaster evolutionary tendency.

Key words: natural disaster, evolution, GERT network, Bayes inference