Tijerina and Garrott (1997) used concepts from reliability theory to develop a model of the potential effectiveness of a crash avoidance system (CAS) to support lane change decisions. In the present paper, a sensitivity analysis is carried out on that model to examine the impact of variations in several model input variables on effectiveness predictions. The variables varied include the detection reliability of the CAS, the proportion of lane change attempts which involve a potential conflict with a principal other vehicle (POV), and the proportions of lane changes executed with each of four prototypical reliability structures. Results of the sensitivity analysis show relative insensitivity to variations in the other input variables if there are only a relatively small proportion of lane changes executed in CAS-only or CAS- driver-in-series fashion. On the other hand, as the proportion of lane changes executed with these two reliability structures increases, effectiveness estimates decrease and negative safety consequences are predicted to grow substantially. Finally, a methodology is presented to characterize the reliability structure of the driver and CAS in an evaluation to further characterize driver behavior with such systems.