In recent years, various safety concepts and innovative ITS technologies have been proposed, developed, and/or implemented in the field aimed at improving roadway safety. To achieve the desired safety benefits while avoiding prohibitive and potentially hazardous field testing, it is critical that proposed safety treatments be extensively evaluated during the design stage and prior to actual deployment. To this end, microsimulation is potentially the most viable tool of choice due to its level of resolution and modeling realism. However, existing microsimulation models are designed to only target normal driver behavior in typical traffic conditions (e.g. either the functional structure or the parameter distributions of these models are deliberately constrained to outlaw unsafe behavior), thus explicitly excluding the occurrence of vehicle crashes. Recognizing the limitations of microsimulation, this research developed an enhanced behavioral car-following model to be implemented in microscopic simulation for facilitating design, testing, and evaluation of safety treatments. Compared to existing models, this new model is built on findings from traffic engineering, human factors, and psychological research, taking into account drivers' perceptual thresholds, perception errors, anisotropy of reaction times, and anticipatory behaviors to allow for vehicle crashes while still capturing typical freeway traffic patterns. High-resolution real-life, crash-inclusive, and crash-free vehicle trajectories collected from the field, in conjunction with aggregated loop detector data, were employed to aid the model development, calibration, and validation. Meanwhile, the project fully used the Minnesota Traffic Observatory I-94 Field Lab, a permanent deployment of sensors and cameras at the area with the highest crash rate in the Twin Cities freeway network. Vehicle trajectories were extracted from real-life crashes collected from a freeway section of I-94WB, making it the first data collection effort that aimed to collect vehicle trajectories from real-life crashes to aid car-following modeling. These data were employed in this study to test, calibrate, and validate the model, resulting in a new model that is successful in replicating these vehicle trajectories as well as crashes.