We present a system-level approach to localizing and tracking users on a basis of different sources of location information. We have applied a combination of the Cricket system and inertial navigation sensors to improve the coverage and accuracy of our framework when used indoors. The system can be applied outdoors as well using, for example, GPS as the source of location data. The overall system architecture is modular and extendible, allowing for creation of location and context aware services and inclusion of additional sources of localization data. Extensive performance evaluation presented shows that the system is both accurate as well as scalable.