A system and method for controlling surveillance cameras is enclosed. In the system, a surveillance camera system includes surveillance cameras configured as VoIP server endpoints. The surveillance cameras receive messages from client devices such as mobile user devices and computer workstations functioning as VoIP client endpoints. The client devices establish a voice communications channel with the surveillance cameras for receiving and controlling one or more streams of image data sent by the surveillance camera over the voice communications channel to the client device. In a preferred embodiment, the voice communications channel includes a control session based on Session Information Protocol (SIP) and a data channel that is based on Real time transfer protocol (RTP). The data channel preferably operates in half duplex mode, which limits the potential for collisions on the data channel when the surveillance cameras are transmitting their image data to the client devices.