The ability of Parkinsonian (PD) patients to control overt visual attention was investigated, by measuring reaction time to a visual stimulus presented at different distances (1.5 deg, 6 deg, and 12 deg) and directions (left or right) from a central fixation point. Prior to the onset of the target stimulus (a square), a cue stimulus appeared just above the fixation point. With equal probability, the arrow pointed to the left, or to the right, or was ambiguous (with two heads). On 20% of their presentations, the left and right arrows pointed in the direction opposite to where the target was to appear. Subjects were informed that 20% of cues would be misleading, and correcting lenses were used to optimise their visual acuity. In previous work with a similar paradigm, only one target eccentricity was used, and subjects were not refracted, leaving open the possibility that PD subjects had more difficulty in seeing the cues and targets. The eight PD subjects had longer reaction times than age-matched normal controls (and were relatively slower for the more eccentric targets), but made fewer errors in all conditions. In particular, they were more accurate than the controls on the presentations when the cue was misleading or ambiguous, suggesting that the PD group were ignoring the cue. It seems unlikely that the subjects discriminate the direction of the cues, given the use of optical correction, and they reported seeing the cues. Our data are consistent with those of other workers who have described a similar ‘disengagement of attention’ in PD (Clark et al, 1989 Neuropsychologia27 131 – 140) and attributed it to decreased catecholaminergic activity following destruction of midbrain structures (Wright et al, 1990 Neuropsychologia28 151 – 159).