This paper proposes a new power control algorithm in multicell networks. It improves the capacity of the cellular network significantly by balancing the utilization of the radio resource and the resulting co-channel interference. In particular, we employ a leader-followers game model to jointly consider the benefits of the users and the network. A global price policy is adopted to optimize the capacity of the whole network by eliminating the users with bad channel conditions. We prove that this game converges to a unique optimal equilibrium in a largescale interference-ideal network. And a price update function is presented to adjust the price adaptively. Simulation results show that this algorithm holds 33-50% capacity gains over the noncooperative game strategy, and achieves comparable performance with the high-complexity centralized geometric programming method even in small-scale networks.