We studied the ability to localize flashed stimuli, using a relative judgment task. Whenobservers are asked to localize the peripheral position of a probe with respect to the midposition of a spatially extended comparison stimulus, they tend to judge the probe as being more toward the periphery than is the midposition of the comparison stimulus. We report seven experiments in which this novel phenomenon was explored. They reveal that the mislocalization occurs only when the probe and the comparison stimulus are presented in succession, independent of whether the probe or the comparison stimulus comes first (Experiment 1).The size of the mislocalization is dependent on the stimulus onset asynchrony (Experiment 2) and on the eccentricity of presentation (Experiment 3). In addition, the illusion also occurs in an absolute judgment task, which links mislocalization with the general tendency to judge peripherally presented stimuli as being more foveal than they actually are (Experiment 4). The last three experiments reveal that relative mislocalization is affected by the amount of spatial extension of the comparison stimulus (Experiment 5) and by its structure (Experiments 6 and 7). This pattern of results allows us to evaluate possible explanations of the illusion and to relate it to comparable tendencies observed in eye movement behavior. It is concluded that the system in charge of the guidance of saccadic eye movements is also the system that provides the metric in perceived visual space.The visual system processes the location of an object as soon as it appears in the visual field. Spatial acuity, measured with various standard methods, is accepted as being of very high precision. It increases from 5 min of arc at 10 0 retinal periphery to better than I min of arc in the fovea. This extremely high acuity is measured with tasks in which the relative position oftwo spatial features is determined (Badcock & Westheimer, 1985;Westheimer, 1981). However, these tasks assess acuity with long-presented, stationary targets with high contrastthus, under optimal viewing conditions. They require only local comparisons of simultaneously presented spatial features, which may not be the best indicators for direct absolute localizations (for an overview, see Skavenski, 1990).We thank Bruce Bridgeman, Asher Cohen, David Rose, and Steven Yantis for helpful comments and suggestions regarding a previous draft of the paper, Birgitt ABfalgand Sonja Stork for carrying out the experiments, and Marina von Bernhardi and Heidi John for stylistic suggestions. Correspondence concerning this article should be sent to 1. Miisseler,