Abstract: 【Purpose/significance】 With the development of digitalization and open innovation, open-source innovation community has become an important form of virtual community. The sharing of project, software code and other knowledge is the key to the good operation of the open-source innovation community, but the knowledge sharing behavior of users in the open-source innovation com⁃ munity is not prominent. It is of great significance to clarify the influencing factors of users′ knowledge sharing in the open-source in⁃ novation community to improve users′ enthusiasm for knowledge sharing and improve the ability of the open-source innovation com⁃ munity.【 Method/process】 Based on the SOR model, taking the content, community and members of the open-source innovation commu⁃ nity as the stimulus factors, and taking the internal characteristics of users′ personal perceived risk and perceived income as the organ⁃ ism, the theoretical framework of the impact factors of users′ knowledge sharing behavior in the open-source innovation community is constructed. According to 329 valid questionnaires, through structural equation model and qualitative comparative analysis of fuzzy sets, the influencing factor path and antecedent configuration are empirically analyzed.【 Result/conclusion】 The research results show that in⁃ tellectual property loss, open source free, incentive mechanism, member trust, perceived risk and perceived income have a significant im⁃ pact on users′ knowledge sharing behavior in the open-source innovation community, and taking community incentives and member trust as the core conditions can form five combined impact paths under the two modes.【 Innovation/limitation】 The research in this pa⁃ per can effectively supplement the existing research in the field of knowledge sharing and knowledge innovation in virtual communities, and provide some reference for achieving better community openness and sharing, as well as knowledge governance and re-innovation. Subsequently, the accuracy of empirical analysis can be improved by combining targeted subjects, primary and secondary data.