Emergencies and their associated negative emotions have a great effect on public health. As a key part of emergency management, government information release (GIR) not only meets the public's health information seeking but also helps to eliminate the breeding and spreading of negative social emotions. From the two aspects of content features and lingual forms, a regression model was built to explore the mechanism of GIR on the regulation of public negative emotions by adopting the theoretical methods of content analysis and emotion calculation. During the emergency outbreak, if the government can timely release information on the incident and respond to the public using rational language, public negative emotions can be alleviated. During the emergency peak, the government should release the event progress, resolution, and disposal information to improve the recognition of public and eliminate negative emotions. According to different stages of emergencies, the government should timely and reasonably utilize the attitude tendency, content type, and lingual form of GIR to effectively regulate the public negative emotions.