The hypothesis that proprioceptively perceived limb position drifts during visual occlusion was re-examined by combining some of the protocols used in previous experiments. Sixteen adult subjects made judgements of static limb position during visual occlusion lasting up to 2 min. In addition, the effect of brief 250 ms "glimpses" of the limb, occasional proprioceptive stimulation and directed attention were examined. Despite, conflicting evidence from earlier experiments, there was clear evidence of a drift in perceived limb position, towards the body, during visual occlusion. This drift was halted if brief glimpses of the limb were provided, or minor re-positioning (without vision) was allowed. In neither case, however, did the supplementary cues reset limb position to its originally perceived position. Drift was amplified when subjects attempted to attend to limb position rather than perform a secondary tracking task. The results are not easily accounted for if drift is considered purely as an effect of peripheral sensor adaptation. A notion of central-drift between visual and proprioceptive maps is suggested as an alternative hypothesis.