The "Landmark Task" is designed to tease apart two major factors in determining line bisection errors in spatial neglect: one whose general nature is perceptual, the other whose nature is motor. On critical test trials, the subject is required to point to whichever end of a mid-transected line is judged as nearer to the transection. Seven out of eight neglect patients pointed consistently to the left end of such lines. Thus their misjudgments were made in the direction opposite to any putative "directional hypokinesia." One patient, however, pointed predominantly rightward on these test trials. Normal controls and unilateral stroke patients were also tested on the Landmark Task. Cueing of one end of a line led to a relative perceptual overestimation of that half of the line in all of these groups.