The parieto-frontal network plays a crucial role in the transformations that convert visual information into motor commands for hand reaching movements. Here we use electroencephalography to determine whether the planning of reaching movements to visual and somatosensory targets involves a similar spatio-temporal pattern of neural activity. Subjects performed reaching movements toward spatial locations defined either by visual (light-emitting diode) or somatosensory (vibration of a fingertip of the contralateral hand) stimuli. To identify the activations associated with sensorimotor transformations, we subtracted the event-related potentials recorded in a "static" task (the stimuli were presented but no movement was initiated) from those recorded in a "reach" task (a reach had to be initiated toward the spatial location of the stimuli). In the visual condition, reach-related activities were observed over parietal, premotor and sensorimotor areas contralateral to the reaching hand. Activation was first observed over parietal areas 140 ms after stimulus onset and progressed to frontal areas. The proprioceptive condition recruited a similar set of structures as for visual targets. However, the temporal pattern of activity within these cortical areas differed greatly. Activity was sustained over premotor and sensorimotor areas throughout the reaction time interval, occurring simultaneously with the parietal activation. These results suggest that a common cortical network serves to transform visual and somatosensory signals into motor commands, but that the interactions between the structures of this network differ. This raises the possibility that different coordinate frames are used to encode the motor error for the two target modalities.