When we move our eyes to examine new regions of interest in a visual scene, object position on the retina shifts and the question is raised whether and how we keep track of an object’s position across a saccade. Verfaillie and colleagues (1997; Verfaillie & De Graef, 2000) demonstrated that transsaccadic memory for object position is poor, but can he improved when a strictly egocentric reference frame is complemented by an allocentric reference frame. When such a reference frame is available, the saccadic status of an object (i.e., saccade source vs. saccade target) appeared to have little effect on the accuracy of transsaccadic position coding, although there was some indication that the saccade target benefited most from allocentric coding. In the present paper, we systematically investigated the impact of allocentric reference frames on the accuracy of transsaccadic position memory for saccade target, saccade source, and saccade bystander. In Experiment 1, the saccade target did indeed benefit most from being integrated in a triangilar object configuration, providing a distinctive allocentric reference frame. In Experiment 2, we determined that this apparent allocentric target superiority was artifactual: When afterimages due to phosphor persistence were eliminated, the saccade source proved to be the object that benefitted most from transsaccadic coding of its position in an allocentric reference frame. Together with the finding that bystander position was also maintained transsaccadically these data challenge theories which assume that the saccade-target region has a privileged role in keeping a transsaccadic record of a scene’s spatial layout.