Mobile user tracking is a major issue in wireless networks. Previous studies and traditional approaches dealt only with tracking algorithms which adapt themselves to the user activity. In this work, we propose a novel approach for user tracking, in which the tracking activity is adapted to both user and system activity. The basic idea is to make the user-location update-rate dependent not only on the user activity (such as the call profile and mobility pattern). Rather, it is also made dependent on the signaling load, which reflects the actual cost of the update operation. Thus, at low-signaling load locations, the users are to transmit location update messages more frequently. To carry out this approach, we propose a load-adaptive threshold scheme (LATS): the network determines for each cell a registration threshold level (which depends on the cell load) and announces it, as a broadcast message, to the users. The user computes its own registration priority and then transmits a registration message only if its priority exceeds the announced threshold level. Thus, whenever the local load on the cell is low, the registration activity increases, while in loaded cells the registration activity decreases. Our analysis shows that the LATS reduces the paging cost, in comparison with other dynamic methods, without increasing the wireless cost of registration. Moreover, if higher user density is coupled with less mobility (e.g., consider vehicles), then the LATS strategy offers further performance improvement. The load-adaptive strategy can be used in addition to any other dynamic tracking strategy. Furthermore, the computational complexity imposed on the user is identical to that required by an equivalent load-insensitive scheme.