The mechanisms underlying visual perceptual stability are usually investigated using voluntary eye movements. In such studies, errors in perceptual stability during saccades and pursuit are commonly interpreted as mismatches between actual eye-position and eye-position signals in the brain. The generality of this interpretation could in principle be tested by investigating spatial localization during reflexive eye movements whose kinematics are very similar to those of voluntary eye movements. Accordingly, in this study, we determined mislocalization of flashed visual targets during optokinetic afternystagmus (OKAN). These eye movements are quite unique in that they occur in complete darkness, and are generated by subcortical control mechanisms. We found that during horizontal OKAN slow-phases subjects mislocalize targets away from the fovea in the horizontal direction. This corresponds to a perceived expansion of visual space and is unlike mislocalization found for any other voluntary or reflexive eye movement. Around the OKAN fast-phases, we found a bias in the direction of the fast-phase prior to its onset and opposite to the fast-phase direction thereafter. Such a biphasic modulation has also been reported in the temporal vicinity of saccades, and during optokinetic nystagmus (OKN). A direct comparison, however, showed that the modulation during OKAN was much larger and occurred earlier relative to fast-phase onset than during OKN. A simple mismatch between the current eye-position and the eye-position signal in the brain is unlikely to explain such disparate results across similar eye movements. Instead, these data support the view that mislocalization arises from errors in eye-centred position information.