[Purpose/significance] The purpose of this paper is to explore the cause and mechanism of social media herd information sharing behavior, and to provide a theoretical reference for the guidance of user information behavior, the supervision of public opinion information and the optimization of network information dissemination mechanism. [Method/process] This paper introduces the variables of self-efficacy in information screening, subjective norms, and information source credibility from the aspects of actors, situations or relationships and objective stimuli, and con? structs a theoretical model by combing the relationship between variables. Then, 262 valid data were collected by ques? tionnaire, and the partial least squares structural equation model (PLS-SEM) was used for data analysis and model veri? fication. [Result/conclusion] Self-efficacy in information screening, subjective norms and information source credibili? ty all have a significant positive impact on herd information sharing behavior of social media users, and in which the im? pact of subjective norms is the greatest. In addition, self-efficacy in information screening can negatively regulate the impact of subjective norms on herd information sharing behavior of social media, but cannot adjust the impact of infor? mation source credibility on herd information sharing behavior of social media. This study reveals the cause and mecha? nism for herd information sharing behavior of social media users, provides a reference for people to understand users' herd information behavior, provides a reasonable explanation for the phenomenon of "screen brushing" in social media, provides an example and scene for the application of triple degree attribution theory, and also provides strategic enlight? enment for service optimization and policy formulation of relevant parties, which has certain theoretical and practical significance.