A method and a device for processing a video conference, including receiving, by a video conference server, first video information sent by a first video terminal and second video information sent by a second video terminal, acquiring, by the video conference server, a preset quantity of personal video information from the received second video information and first video information, and synthesizing the preset quantity of the personal video information to generate a synthesized video, and sending, by the video conference server, the synthesized video to a third video terminal to display. The method and device for processing a video conference avoid a poor display effect due to site disparity when the integrated video of the site is used as a basis, and provide flexibility for maximizing the information display efficiency.