Cooperative collision warning (CCW) systems use communication networks as a main component for creating situational awareness and eventually hazard detection. Simulation and analysis of such systems are generally more complicated due to the system being composed of components from very different domains of communication and vehicle safety. These components are inherently developed and modeled in different domains, as their basic operations are usually defined and engineered by researchers from different disciplines. Creating a simulation tool for CCW systems requires combining simulation models that are developed using different methodologies. As a result, a unified tool for study of such systems is not readily available. In this paper, we describe a co-simulation tool that models both components of communication and hazard prediction in one framework. The tool uses several different levels of abstraction for the communication model, while modeling the application in a precise manner. We show that communication component can indeed be abstracted to some level without loss of accuracy. We also investigate the impact of using different communication strategies on the performance of CCW systems, and demonstrate the possibilities that the co-simulation tool offers.