The study tested the hypothesis that movement and self-referent information differ, and that children can use both to code unseen locations. In one experiment blindfolded sighted children made more errors with changed than with unchanged accessing movements, despite unchanged reference conditions. Rotation decreased accuracy further, although most responses were to the correct side. Age did not interact with experimental conditions. In another experiment blind children and matched blindfolded controls were less accurate with changed movements; and rotations were more difficult still. Visual experience, but not age, interacted with conditions. The blind showed significantly more self-reference (responses to the original side in rotations), and were somewhat worse also with changed movements alone but did not differ from controls when movements and references were unchanged. The results support the hypothesis that movement and self-referent information differ. Visuospatial experience and task conditions rather than age seem to determine the extent to which children use either form of information.