We have used the technique of PET to chart the cortical areas activated by visual motion in the brain of a patient with a severe impairment in the ability to recognize the motion of objects (akinetopsia), following bilateral lesions which have so far been presumed to include area V5. High resolution MRI of her brain showed that the zone occupied by area V5 had indeed been destroyed bilaterally. Positron emission tomography activation images, co-registered to the MRIs, showed three principal regions of the cortex activated by motion. These were located (i) bilaterally in the precuneus of superior parietal cortex (area 7 of Brodmann); (ii) bilaterally in the cuneus (a region considered to represent upper V3); (iii) in the left lingual and fusiform gyri (possibly lower V3 and adjacent areas). In contrast to normal subjects, there was no significant activation of area V1 or V2. The stimuli used for scanning were chosen by prior testing of the patient's visual capacities. The control stimulus was a static random distribution of light squares on a dark background. In the moving stimulus these squares moved coherently, the direction of motion changing periodically between the cardinal directions (left, right, up and down). It was ascertained that the patient could correctly identify these directions. We also found (i) that her occasional errors were always in the direction opposite to the motion presented, so that her identification of axis of motion (i.e. vertical or horizontal) was 100% correct; (ii) that when a few static squares were added to the moving the display her identification of direction fell to chance but her identification of the axis of motion remained 100%; (iii) that when a few squares moving opposite and orthogonal to the predominant direction of motion were incorporated, her performance on both direction and axis fell to chance; (iv) that she was unable to identify motion in oblique directions between the horizontal or vertical axes, always guessing one of the cardinal directions. In accounting for her residual vision in terms of cortex, which remains active, we hypothesize; (i) that the bilateral loss of V5 has affected direction sensitive mechanisms at other sites in the cortex which are interconnected with V5 and (ii) that in consequence her performance on our tests reflects the properties of dynamic orientation selective mechanisms that were also differentially activated by the stimuli used during scanning.