Location management is one of the most important issues in mobile networking systems. Most current mobile systems employ a two-tier database structure to store mobile users? information. HLR (Home Location Register) is used to retrieve user?s information, and VLR (Visitor Location Register) is the temporal database for storing visitors? profile. In the case that the number of users exceeds the capacity of the VLR due to mobility, database overflow occurs. In this paper, we propose a new policy, the least frequent replacement policy, which selects the least frequently accessed users as replacement candidates. Extensive simulation is carried out for predicting the performance of the existing replacement policies and the proposed least frequent policy. Our simulation results show that the most idle and least frequent policies outperform the inactive and random replacement policies in most cases and that the least frequent policy performs slightly better than the most idle replacement policy.