In a Voice over Internet Protocol (VoIP) terminal supporting an optimal intercom service and a session connecting method thereof, a VoIP terminal announcement module of the VoIP terminal performs an announcement process by transmitting address information of the VoIP terminal to other terminals located in a same network, and acquires address information of said other terminals. If the address information acquired by the VoIP terminal announcement module includes information of a counterpart terminal called by a user, a call controller sends a session connecting invite message directly to said counterpart terminal using the acquired address information of the counterpart terminal. As a result, it is possible to provide intercom service to a user without going through an external SIP proxy server.