Cooperative vehicle safety systems rely on periodic broadcast of each vehicle’s state information to track neighbors’ positions and therefore to predict potential collisions. The most pressing challenge in such systems is to maintain real-time tracking accuracy while avoiding channel congestion. Previous work on congestion control in CVSSs primarily focuses on the effort of individual vehicle to adjust the contention window based on its local channel status, which may cause the channel resource to be monopolized by a small number of vehicles. As a result, unfairness issue arises. To overcome this problem, in this paper, we present a game-based channel access approach for contention window adjustment. In this approach, a utility function that defines the relationship between vehicle tracking accuracy and throughput is firstly proposed. Then, based on the utility function, a non-cooperative dynamic game model with a penalty function is established. In order to implement the game model on each vehicle, a distributed channel access algorithm is proposed. This algorithm adjusts the contention window for each vehicle, and hence, guides each vehicle to the Nash equilibrium. Simulation results confirm that the gamebased channel access approach can guarantee a fair share of channel resources while achieving the optimal tracking performance for vehicles.