The continuing explosion in the amount of readily available spatial data has given rise to significant problems in accessing the most task-relevant spatial information. At the same time, advances in mobile computing have yielded handheld platforms powerful enough to deliver spatial information interactions, but with usability issues in interfacing and wireless data transmission. We are developing techniques to manage information overload in mobile interactions with spatial knowledge repositories. In this paper we outline a novel application called CoMPASS (Combining Mobile Personalized Applications with Spatial Services) that incorporates knowledge management and spatial modeling approaches to provide end-users with spatial data tailored to their immediate requirements and long-term spatial preferences. CoMPASS presents spatial data in the form of personalized vector maps, based on user models, which enhance both application quality and efficiency by delivering the most relevant information and map content.