A surveillance video accumulation system pertaining to the present invention includes a plurality of surveillance camera units and a video accumulation unit that are connected with each other via a network. Among pieces of video data that have been encoded according to a prediction encoding method and transmitted from all the surveillance camera units, the video accumulation unit accumulates video data transmitted from a selected one of the surveillance camera units. Before the selection, if specified as a surveillance camera unit that is expected to shoot a subject, each of the surveillance camera units encodes video data of the subject based only on intra frame prediction and transmits the encoded video data to the accumulation unit.