An image display apparatus is disclosed in which the position of an exit pupil can follow the movement of a pupil of an observer even with a small size. The apparatus includes an image-forming element which forms an image, an ocular unit which causes light from the image-forming element to form an exit pupil, and an optical element which reflects the light from the image-forming element toward the ocular unit. The optical element has an optical power and is rotated about a focal point of the ocular unit on the side closer to the image-forming element to move the exit pupil.