Three experiments are reported examining judgements of the centre of a stick in a patient with unilateral neglect after right hemisphere damage. Replicating previous data [35, 37], judgements showed more evidence of neglect when pointing rather than when a grasp response was used (Experiment 1), particularly when pointing preceded grasp (Experiment 2). Neglect also increased for longer sticks and when sticks fell in the patient's left hemispace; the effects of stick length and hemispace were additive with those of response (point vs grasp). Experiment 3 showed that the advantage for grasp over pointing responses occurred only when performance was guided by on-line visual feedback, and it emerged only during the end part of the reach trajectory. The results are discussed in relation to the role of visual feedback in movement control.