This invention discloses a multi-point video conference system and the media processing method thereof. The multi-point video conference system includes a softswitch, an application server and terminals. The softswitch is responsible for protocol adaptation and calling process, and performing the interconnection with other systems as an external interface of the conference system. The application server is responsible for service logic control and charging. The characteristic of this invention lies in that: the terminals behaved as conference participants including a conference chairman and ordinary conference members. Any terminal registered on the softswitch could be a conference member, and the conference chairman must be a SIP soft terminal with video conference service capability in order to realize the media control. The invention realizes the media control on the basis of the soft terminal behaved as the conference chairman, therefore any terminal user that has registered on the softswitch can initiate or participate a multi-point video conference, and the participant terminal number of the video conference is not limited.