A multi-view video encoding apparatus and a method thereof are provided. The apparatus includes: a multi-view video input unit which receives a multi-view video; an I frame location determination unit which determines a location of an I frame which is a basis for encoding the multi-view video according to characteristics of multi-view frames at a predetermined time; and an encoding unit which encodes the multi-view video based on the determined location of the I frame. When the apparatus and the method are used, the location of the I frame, which is a basis for compression-encoding a multi-view video, can be adaptively determined and not fixed, thereby enhancing prediction efficiency and compression rates.