Vehicle interactions with roadside restraints such as crash barriers and bridge parapets are extremely complex events and the outcome can be greatly influenced by very local effects. Such impacts typically take place over a time period of 2–3 seconds which presents particular challenges when simulating the process numerically. The inclusion of these local effects necessitates very small time increments during the analysis and this consequently results in extremely long analysis times.A technique has been developed that predicts the outcome of vehicle to roadside restraints in a computationally efficient manner. The methodology capitalises on advanced features and material models available in the LS-DYNA finite element code, and avoids the use of detailed vehicle models that can over-complicate the analysis and lead to vehicle-specific performance predictions for the barrier. The technique uses advanced beam element and contact formulations and makes use of a generic vehicle representation. Use of a generic vehicle in this form removes the possibility of vehicle-specific performance predictions for the barrier, and allows the computational resources to be focussed on analysis of the barrier rather than the vehicle, which is effectively only a loading mechanism for the barrier.The approach has been validated throughout against test results and has been proved capable of predicting all conventional roadside restraint performance measures including injury measures as defined in BS EN 1317. Validation has taken place against a range of restraint systems including aluminium parapets, reinforced concrete parapets and wire-rope safety fences, which makes the approach an effective tool for establishing legislative test conditions and for researching and predicting RRS performance.