A coding apparatus includes: a storage section that stores information for coding an image area to be currently coded in a direct prediction mode that predicts motion information of an image area to be coded by referring to motion information of a previously coded image area; a deriving section that derives motion information of the image area in the direct prediction mode by referring to the information stored in the storage section; a selecting section that selects a prediction mode for coding the image area to be coded, in accordance with the motion information derived by the deriving section; and a determining section that determines information necessary for the deriving section to derive motion information of an uncoded image area, from coding information applied in the prediction mode selected by the selecting section to the image area to be coded, and stores only information determined necessary into the storage section.