Abstract We perceive scenes as stable even when eye movements induce retinal motion, for example during pursuit of a moving object. Mechanisms mediating perceptual stability have primarily been examined in motion regions of the dorsal visual pathway. Here we examined whether motion responses in human scene regions are encoded in eye- or world centered reference frames. We recorded brain responses in human participants using fMRI while they performed a well-controlled visual pursuit paradigm previously used to examine dorsal motion regions. In addition, we examined effects of content by using either natural scenes or their Fourier scrambles. We found that parahippocampal place area (PPA) responded to motion only in world- but not in eye-centered coordinates, regardless of scene content. The occipital place area (OPA) responded to both, objective and retinal motion equally, and retrosplenial cortex (RSC) had no motion responses but responded to pursuit. Only PPA’s objective motion responses were higher during scenes than scrambled images, although there was a similar trend in OPA. These results indicate a special role of PPA in representing its content in real-world coordinates. Our results question a strict subdivision of dorsal “what” and ventral “where” streams, and suggest a role of PPA in contributing to perceptual stability.