In recent years, there has been a tremendous increase in personal media data stored on people’s PCs, mobile devices, and social media sites in the web. Additionally, people are increasingly collaborating and interacting by sharing and commenting media items. These trends call for retrieval services integrating resources heterogeneous in update frequencies, media types, and size. In this context, peer-to-peer (P2P) technologies offer interesting solutions. When performing a query on certain types of P2P networks, resource selection is important. Compact summaries (i.e. resource descriptions) of each peer’s data collection are known to other peers and used by them in order to determine promising peers for a given query. Summaries have to describe not only textual information, content-based media features, and information about date and time, but also the locations where e.g. images were taken, videos were recorded, or to which a user travelled. The present paper proposes and evaluates different resource selection techniques based on descriptions of the geographic footprint of personal media archives when querying for media items that are geographically close to a given query location. These techniques are not restricted to P2P networks and can e.g. be applied in hybrid index structures or distributed IR systems in general.