The role of afferent inflow and efferent outflow (or command) signals in judgements of limb position has been debated for over a century. One way to assess this is to check for changes during complete paralysis, with the current view being that perceived movements or position changes do not usually accompany attempts to contract paralysed muscles. To re-examine this, we asked six naïve subjects to carry out a simple position-matching task at the wrist. In the absence of vision, subjects accurately perceived the position to which their right wrist had been moved by the experimenter by matching it with their left hand. There was no significant change in perception when position was matched during sustained flexion or extension efforts. Then we paralysed and anaesthetized the right arm with ischaemia in order to produce a ‘phantom’ hand. The perceived position of the wrist changed by more than 20 deg when subjects attempted to flex or extend their hand when it was paralysed and anaesthetized. Further studies showed that this illusion was not dependent on the way in which the paralysis was produced and that the size of the position illusion increased when the level of effort during paralysis increased. These results establish for the first time a definitive role for ‘outflow’ signals in position sense.