This study investigated the role of vision in the encoding and reproduction of movement. Kinaesthetic reproductions of a kinaesthetically presented two-dimensional movement were compared with reproductions where vision was present either during the standard or at reproduction. The main finding was that the presence of vision during the standard resulted in poorer accuracy and greater underestimation of movement size than when it was absent throughout or present during reproduction. The presence of vision during the standard, however, resulted in less distortion of the linear components of movement shape although no such effect was found for the angular components. The initial direction of movement was reproduced more accurately where visual experience of the movement was given, either during the standard or during reproduction.