We sought to determine the effects of prism adaptation on peripherally cued visual attention shifting in patients with spatial neglect, using a task devised by Egly et al. (J Exp Psychol Gen 123:161-177, 1994) based on the classic Posner paradigm. This task allowed a comparison of "within-object" versus "between-object" attention shifts. A display was presented containing two parallel outline rectangles, and subjects were asked to make rapid responses to a target, which would appear at one end of one of the rectangles. The target location was pre-cued with 75% validity: on invalid trials attention was directed either to the other end of the same rectangle, or to the other rectangle. Healthy subjects and right-hemisphere patients without neglect showed a left-right symmetrical pattern, with a larger validity effect when required to shift attention between rectangles, thus indicating a greater difficulty of attention-shifting between than within the respective shapes. The neglect patients showed the typical leftward "disengage deficit" previously observed in neglect, but only for attention shifts between objects, indicating that the effect is object-based rather than purely spatial. A comparison of vertical and horizontal shift costs showed that this attention-shifting deficit for left-hemifield target stimuli was directional rather than hemifield-based: it was absent for vertical shifts of attention within the left hemifield. Finally, we found that prism adaptation abolished the disengage deficit. We found no effects of prism adaptation in the control subjects. We argue that prism adaptation has a powerful effect on one of the fundamental manifestations of the neglect syndrome.