This paper presents research of developing a new behavioral car-following model that is pertinent to the true nature of "less-than-perfect" human driving. Unlike traditional car-following models that deliberately prohibit vehicle collisions, the proposed model tries to incorporate the weakness and risks associated with everyday driving and strives to capture both safe as well as unsafe driving behaviors. It is important to note that parameters of this model have direct physical and behavioral meaning; this allows that vehicle collisions, if any, are replicated as a result of behavioral patterns rather than simply being numerical artifacts. Vehicle trajectories extracted from real-life crashes were employed to test the model's capability of replicating freeway rear-end collisions. Furthermore, RTKGPS crash-free trajectory data were used to validate the model against normal driving behavior. Testing results indicate that the proposed model is able to emulate both normal as well as unsafe driving behaviors that could lead to vehicle collisions. Finally, feasibility of integrating the proposed model with existing microsimulators is briefly discussed. Eventually this will facilitate studying crash mechanisms at a high definition microscopic level as well as enabling safety-related evaluation using microsimulation tools.