Agent-based simulation may provide an effective method of evaluating safety issues in air traffic control. For example, simulation of human performance models of controllers and pilots, interacting with aircraft avionics systems and with communication, navigation, and surveillance technologies, may assess both the individual behavior of these agents and the overall emergent behavior of the air traffic control system. This paper describes a collaborative effort in developing an agent-based simulation of air traffic control. In order to exercise the capabilities of the proposed approach, a test scenario was developed to investigate the impact of new avionics systems which provide alerts of clear air turbulence (CAT) to flight crews. An agent-based simulation using the Reconfigurable Flight Simulator (RFS) software architecture and its asynchronous timing methods is outlined. Adaptation of the Man-machine Integration Design and Analysis System (MIDAS) human performance model to model the behavior of pilots and air traffic controllers and to interact with RFS is detailed, including description of an algorithm implemented to synchronize the timing between RFS and MIDAS. This paper then discusses the use of agent-based simulation incorporating human performance models as a means to assess safety.