We report a follow-up study on seven arm amputees in whom magnetic source imaging had originally revealed a strong correlation between the amount of cortical invasion of the deafferented cortex and the amount of pain evoked sensation mislocalized to the phantom limb. This re-examination was performed in order to corroborate the phenomenon of mislocalization. On follow-up examination for mislocalization 4 weeks later, a close correlation had remained between the original amount of cortical representational reorganization of the amputation zone (at the first examination) and the number of sites from where painful stimuli evoked sensations referred to the phantom limb, i.e. the amount of perceptual mislocalization, at the second examination. However, contrary to our expectation, the topography of referred sensation had completely changed in every patient. These results suggest that while the overall extent of reorganization is a rather stable phenomenon, the concomitant changes in the pattern of sensory processing are not. This may be due to the fact that alterations of sensory processing are not hardwired, but are rather mediated by an extensive and interconnected neural network with fluctuating synaptic strengths. This mechanism may be of importance for neurological rehabilitation.