In this paper, a new method of dynamic spectrum sharing with bidding is proposed in cognitive radio system. The Cournot game model of the proposed method is established, and it is proved that Nash Equilibrium of the new game model is existed, which can be found out by eliminating inferior point repeatedly. The method takes both primary and secondary users' earnings into account respectively when sharing the spectrum. We consider the problem of spectrum sharing among a primary user system and multiple secondary users. With the aim of maximizing its own profit, the spectrum market server of primary user decides how many spectrums to be lent based on the price each secondary user pay for the spectrum. With the game model, secondary users bid against each other to maximize every secondary user's revenue, and achieve the Nash Equilibrium finally. The simulation results show that this spectrum sharing algorithm under the bidding mechanism has fast convergence, and can achieve the Nash equilibrium finally. Compared with no bidding mechanism, the can make much more profit for the primary users.