A growing body of evidence demonstrates that vision for perception and vision for action are mediated by separate neural mechanisms. After briefly reviewing the neuropsychological evidence for this division of labor in the human visual system, we explore the evidence for a dissociation between perception and action in neurologically intact individuals. A number of studies have shown that unseen visual events can sometimes elicit movements of the hand and limb, despite the fact that subjects have no visual phenomenology of those events. Other work has shown that perceptual judgements about the location and size of objects can be quite different from the scaling of skilled actions directed at those objects. For example, size-contrast illusions, such as the Ebbinghaus illusion, have been shown to have little effect on the scaling of the grasp. Similar dissociations have been demonstrated in other studies in which psychophysical judgements about the dimensions of objects in the far peripheral field bear little relation to the calibration of grasping movements directed at those objects. Together with the neuropsychological work (and neurophysiological studies in the monkey), these findings provide compelling evidence for the operation of separate visual mechanisms in everyday life. In other words, what we think we see is not always what guides our actions.