The implementation of data warehouses (DWs) for the huge databases has become one of the major needs of the current times. Large organizations having different small databases need a proper integration of their small databases and development of data warehouses. The development of data warehouse for an organization helps in storing the statistical information that can be helpful in decision making process later on. This statistical information needs to be stored with proper security and confidentiality. At the same time, this information also needs to be readily available to all authorized users, in order to ensure the accuracy and consistency of data. Though, so far, a lot of work has been done towards the development of data warehouses, but not much attention has been given towards improved and optimal implementation of security measures in data warehouses. This paper highlights some of the security approaches, like classical security models and implementation of filter over data before they are kept in statistical form in a data warehouse. The comparison of these approaches and introduction of a hybrid approach of the security implementation in data warehouses is the main focus of this paper. © 2010 IEEE.