How to identify influential nodes in complex networks is still a hot topic.In order to solve the issue of identification node importance in complex network，a new method is proposed based on belief function.Some background knowledge of belief function，complex network and algorithms of identification vital nodes are recalled，and a model to identify node importance is built.A framework of discernment is built.Some attributes and values of nodes are adopted，which will be transformed into the styles of belief function under the same framework.And then，the Dempster combination rule is carried out to combine these belief functions.After that，a comprehensive belief function of node is generated，and a method is used to transform it into a single value.The node is ranked based on the single values.Numerical example indicates that some shortcomings of mentioned simple algorithms of node importance are overcome by new proposed model.The new proposed method is of effectiveness and rationality，and it can be generalized.