[Purpose/significance] Research on the characteristics and influencing factors of browsing behavior of research publications in academic social networks can help users improve the dissemination effect and influence of research publications based on academic social networks. [Method/process] With 32 501 research publications selecting from ResearchGate, this paper reveals the overall distribution of reads and the time-series change in weeks. Fromthe four dimensions of first author, number of collaborators, presentation form and quality level, this paper uses correlation analysis combined with principal components analysis and logistic regression analysis to explore the influencingfactors of research publications’reads. [Result/conclusion] The research discovers that the reads of research publications in academic social networks present the power-law distribution, and the dissemination effect of most researchpublications is still small in weeks. According to its time-series change, research publications can be divided into fourtypes: swell and swell, continuous ascent, steady change and low-read. Full article and abstract pictures reflecting thecompleteness of research publications’content on RG, the average reads of the first author, the research interest and citations reflecting the quality of research publications all have significant positive impacts on the browsing level of research publications, and the degree of impact decreases in turn. Enriching the display form and content of research publications and improving the visibility of complete publications are the keys to improve the reads of research publications for academic social networks and their users.