In order to effectively assess all types of security risks in the important event, the important decision-making basis for security risk warning and emergency management of important event is provided by analyzing the coupling relationship and evolution mechanism between various risks. The criminal causes, management defects, security and emergency system construction are analyzed from the possibility of accidents and risks. The multi-source data risk assessment system based on five subsystems and its index set of human factors, management factors, site factors, event factors and audit factors are proposed. The weight of each index in the assessment system is determined by the method of information entropy, and then the risk grade of important event is determined according to the weight calculation function and the improved fuzzy matter element model. The verification with an example shows that: the risk assessment model is optimized by combining entropy weight with fuzzy matter-element model，the influence of weight data extreme value was weakened，the qualitative description and quantitative analysis of multi-source data could be combined，and the subjective error was reduced. The risk grade of important event is reasonably evaluated, and the assessment effect is basically consistent with the expert inspection analysis, which shows that the method had certain application value.