We studied a patient (J.J.) with bilateral damage to those regions of the posterior parietal cortex (PPC) thought to be involved in prism adaptation. We demonstrated, for the first time in a parietal patient, that J.J. was unable to adapt to the visual perturbation induced by the optical prisms with either hand within four times the number of trials required by healthy adult subjects. We offer a novel account for the role of the PPC in prism adaptation: that the reach direction to the veridical target location specified in extrinsic limb-based coordinates must be de-coupled from the gaze direction to the perceived target location specified by intrinsic oculocentric coordinates in order to produce spatially accurate movements. This spatial discrepancy between gaze direction and reach direction may provide the necessary training signal required by the cerebellum to update the current internal model used to maintain spatial congruency between visual and proprioceptive maps of peripersonal space. The hypothesis is discussed in relation to recent disconnectionist accounts of optic ataxia.