Provided is a social network initial key node selection method considering a delay constraint. The method comprises the steps that influence among users in a network is indicated as the credit value size accumulated on nodes after the credit allocation process is finished in a quantized mode by combing the meeting probability and the activation probability; by means of study on a network structure and user behavior records, the transmission block action which is produced by trying to meet and activate other users is converted into the spreading increment path metric, and credit is allocated in a limiting mode by combining delay constraint conditions; the nodes of which the marginal revenue is the maximum are selected through a greedy thought in a recursive mode to compose an initial key node set finally. According to the social network initial key node selection method considering the delay constraint, the novel social network initial key node selection method is created; by means of the combination of probability events and the delay constraint conditions, the defect that a rule of evaluating the note influence only on the basis of node degree values is improved, the execution efficiency in the selection process is improved, and the communication process of behaviors in the network is truly and effectively simulated and predicted.