The disclosed image encoding method includes: a first step in which a motion reference block spatially adjacent above an encoding target block, and motion reference blocks adjacent to the left and right of the encoding target block, are selected from among encoded pixel blocks to which inter-prediction has been applied; a second step in which usable blocks that contain mutually different motion information are selected from among the motion reference blocks, said motion information being candidate motion information for application to the encoding target block; and a third step in which one selected block is selected from among the usable blocks. The disclosed image encoding method also includes: a fourth step in which the motion information in the selected block is used to generate a predicted image for the encoding target block; a fifth step in which a prediction difference between the aforementioned predicted image and the original image is encoded; and a sixth step in which selection information which identifies the selected block is encoded via reference to an encoding table predetermined in accordance with the number of usable blocks.
