A call admission control (CAC) scheme used in hierarchical mobile IPv6 (HMIPv6) network is proposed, which solves the ideal call number of each cell adaptively to maximize system’s profits, under the restriction of system’s overload probability and capacity and in terms of hosts’ mobility characteristics and call characteristics. The CAC threshold of each cell used in making decisions of accepting or rejecting a new call by network is solved by establishing a Markov model in HMIPv6 network, which can make system’s call number up to the ideal value. Simulation analyzes the factors affecting the optimal regional size, providing a reference for designing a dynamic regional mobility management scheme. The simulation analyses reveal that with the increase of calls’ handoff probability and average duration time, the optimal regional size increases while the regional ideal call number decreases; and with the restriction of overload probability relaxes, the new call blocking probability (CBP) comes down while the handoff dropping probability (HDP) goes up. Finally, a performance comparison result between the proposed scheme and the mobile IP reservation (MIR) scheme demonstrates that the average values of CBP and HDP in the proposed scheme are lower than those in MIR during 2000 simulation intervals, which means the optimization is realized.