When a major public health incident breaks out, in order to prevent the explosion of multiple types of public opinion, relevant government departments need to guide the online public opinion according to the needs and characteristics of different audiences in order to achieve reasonable regulation and control. In this process, gender differences among the participating public in areas such as comprehension ability often affect the effectiveness of government guidance. A proper understanding of these differences will enable the government to allocate resources on the basis of needs to save resources and achieve the same goals with half the effort. This paper takes the outbreak of the COVID-19 as an example to analyze the gender differences among users in terms of the overall volume of participation and specific participation behaviors from the dimension of time and geographical locations. A total of 735,271 comments posted by users in responding to tweets published by 144 official government accounts on Weibo during the COVID-19 outbreak were collected and analyzed with a combination of the methods of natural language processing and propensity score analysis. The results show that in comparison to male users, female users participated more, and their responses were more emotionally expressive. Female users tended to respond faster than male users by 30 minutes to an hour, which allowed female users to play a more important role in the process of government guidance of public opinion during major public health incidents. Therefore, this study further provides policy recommendations for the government to provide reasonable guidance of public opinion and give future direction.