Abstract: 【Purpose/significance】The frequent occurrence of online public opinion reversal in emergencies has brought severe chal⁃ lenges for relevant public opinion management departments to conduct public opinion governance. Mastering the formation path of on⁃ line public opinion reversal can effectively improve the predictability of the reversal of public opinion and help them timely implement the supervision and guidance of online public opinion.【 Method/process】Based on the perspective of symbiosis theory, this paper con⁃ structs a path analysis framework for the formation of online public opinion reversal in emergencies. 30 public opinion cases of hot emergencies are taken as research samples, and fuzzy set qualitative comparative analysis (fsQCA) is used to extract seven conditional variables, including media coverage, Internet user performance, content sensitivity, information transmission characteristics, cognitive bias, group pressure and government intervention. Explore the combination logic and formation path of network public opinion reversal. 【Result/conclusion】The results show that the formation of network public opinion inversion tends to be influenced by multivariate con⁃ figuration. According to the symbiotic unit, the reverse path of network public opinion is summarized into three combinatorial paths: the contradiction type, the deviation type of media report and the co-action type, and the multiple factors interact with each other. Jointly pro⁃ mote the phenomenon of the reversal of public opinion on the Internet【. Innovation/limitation】Revealing the complex causal relationships between various conditional variables in the formation path of online public opinion reversal effectively expands the research dimension of symbiosis theory. In the future, the capacity for case selection can be increased to verify the universality of research results.