To reduce an overall code amount by much more compressing encoded information in an image encoding system employing hierarchical encoding.

In an image encoding device for encoding an image while using a motion vector for the unit of a block resulting from dividing a picture of each layer while using scalable encoding for hierarchically encoding an image, a predictive motion vector candidate generation section 141 derives candidates of a predictive motion vector from motion vectors of a predictive block subjected to encoding in an extension layer and of an encoded predictive block proximate to the predictive block subjected to encoding in the same picture and derives candidates of a predictive motion vector from the motion vector of the predictive block in an encoded picture in a lower layer than the predictive block subjected to encoding. A predictive motion vector selection section 146 determines a candidate of the predictive motion vector to be the predictive motion vector of the predictive block subjected to encoded from among the derived candidates of the predictive motion vector.