In this age of information, the efficient use of electronic communications is essential. As technology advances and becomes more complex, it is imperative that groups be able to discuss ideas and disseminate information among members effectively. Multicast groups are established to facilitate these information transactions. Since the members of these groups may be spread across the globe, the communications must be secure as well as efficient. Secure multicasting is an active area of research today. Though the areas of secure multicast group architecture, key distribution, and sender authentication are under scrutiny, one topic that has not been explored is how to integrate these with multilevel security. Multilevel security is the ability to distinguish subjects according to classification levels, which determines to what degree they can access confidential objects. In the case of groups, this means that some members can exchange messages at a higher sensitivity level than others. The Bell-La Padula model outlines the rules of these multilevel accesses. In multicast groups that employ multilevel security, some of these rules are not desirable so a modified set of rules was developed and is termed differential security. This thesis proposes three possible methods in which to set up a differenti0y secure multicast group: a naive approach, a multiple tree differential security (DiffSec) approach, and a single DiffSec tree approach. In order to evaluate the performances (in terms of the number of links used per packet transmitted) of these approaches, extensive simulation experiments were conducted by varying the network connectivity and group size for both uniform and nonuniform membership distribution across security levels. Our studies show that the multiple tree and single DiffSec tree approaches perform much better than the naive situation. While the multiple tree approach could be implemented using current technology, this scheme consumes many times more addresses and network resources than the single DiffSec tree approach. From our studies, we conclude that the single DiffSec tree is a viable option for supporting multilevel security as it maximizes the resource utilization and is also scalable.