Objective: To investigate cerebral activity associated with allodynia in patients with neuropathic pain. Methods: The brain responses of 27 patients with peripheral (5), spinal (3), brainstem (4), thalamic (5), lenticular (5), or cortical (5) lesions were studied with fMRI as innocuous mechanical stimuli were addressed to either the allodynic territory or the homologous contralateral region. Results: When applied to the normal side, brush and cold rubbing stimuli did not evoke pain and activated a somatosensory “ control” network including contralateral primary (SI) and secondary (SII) somatosensory cortices and insular regions. The same stimuli became severely painful when applied to the allodynic side and activated regions in the contralateral hemisphere that mirrored the “ control” network, with, however, lesser activation of the SII and insular cortices. Increased activation volumes were found in contralateral SI and primary motor cortex (MI). Whereas ipsilateral responses appeared very small and restricted after control stimuli, they represented the most salient effect of allodynia and were observed mainly in the ipsilateral parietal operculum (SII), SI, and insula. Allodynic stimuli also recruited additional responses in motor/premotor areas (MI, supplementary motor area), in regions involved in spatial attention (posterior parietal cortices), and in regions linking attention and motor control (mid- anterior cingulate cortex). Conclusion: On a background of deafferentation in the hemisphere contralateral to stimuli, enhanced or additional responses to innocuous stimuli in the ipsilateral hemisphere may contribute to the shift of perception from innocuous toward painful and ill- defined sensations. Objective: To investigate cerebral activity associated with allodynia in patients with neuropathic pain. Methods: The brain responses of 27 patients with peripheral (5), spinal (3), brainstem (4), thalamic (5), lenticular Results: When applied to the normal side, brush and cold rubbing stimuli did not evoke pain and activated a somatosensory “control ”The same stimuli became severely painful when applied to the allodynic side and the activated regions in the contralateral hemisphere that mirrored the“ control ”network, with, however, lesser activation of the SII and insular cortices. Increased activation volumes were found in contralateral SI and primary motor cortex (MI). Whereas ipsilateral responses had very small and restricted after control stimuli, they represented the most salient effect of allodynia and were observed mainly in the ipsilateral parietal operculum (SII), SI, and insula. Allodynic stimuli also recruited additional responses in motor / premotor areas (MI, supplementary motor area), in regions involved in spatial attention (posterior parietal cortices), and in regions linking attention and motor control (mid-anterior cingulate cortex). Conclusion: On a background of deafferentation in the hemisphere contralateral to stimuli, enhanced or additional responses to innocuous stimuli in the ipsilateral hemisphere may contribute to the shift of perception from innocuous toward painful and ill-defined sensations.