Abstract: Dynamic channel sharing policy is the key technology to improve the spectrum utilization in cognitive radio networks. First, the scenario is considered that several cognitive networks with finite user population coexist. Then, a three-dimensional Markov chain model is developed to analyze the performance of the partial channel sharing policy in the cases with and without channel handoff. Finally, the blocking probability, the forced termination probability, handoff probability of the users and the system throughput are analyzed by comparing with the static spectrum allocation policy and the hierarchical sharing policy. The simulation results show that the system with partial channel sharing policy will gain higher throughput by tolerating lower handoff probability and forced termination probability.