Roadside barrier systems play an important role in reducing the number of fatalities and the severity of injuries in highway crashes. After decades of work by researchers and engineers, roadside barriers have been improved and are generally effective in preventing head-on collisions and thus crash fatalities. To further improve the performance of highway safety devices and develop new systems, a good understanding of occupant injuries is required. Although incorporating occupant responses and/or injuries into the design of safety devices is highly recommended by the current safety regulations, there are currently no studies that can be used to develop official guidelines or standards. Despite its usefulness in understanding the crash mechanism and improving vehicle crashworthiness, crash testing is very expensive and restricted by the crash scenarios that can be investigated. In addition, no crash test dummy is incorporated in majority of the crash testing of roadside barriers.With the recent advances in high performance computing and numerical codes, computer modeling and simulation are playing an important role in crash analysis and roadside safety research. In this study, the finite element model of a Hybrid III 50th percentile male dummy was developed for studying the driver's responses in vehicular crashes into highway barriers. After validation by standard crash tests, the dummy model was combined with the finite element model of a 2006 Ford F250 pickup truck and used in simulations of the vehicle impacting a concrete barrier and a W-beam guiderail under different impact speeds and angles. Finally, the dummy responses in these simulations were analyzed by correlating with existing human injury criteria so as to correlate impact severity to vehicular responses and ultimately to barrier performances.