The ability to quickly design new vehicles with optimal crashworthiness has long been a goal of automotive manufacturers and Tier 1 suppliers alike. This paper takes steps towards that goal by automating manual design iterations. The crashworthiness of an instrument panel was optimized using LS-OPT.In one design experiment, optimizing the gauges of non-styled parts in the instrument panel reduced the simulated force in a Bendix test setup by around 30%. In a second design experiment, optimizing the shape of non-styled parts in the instrument panel with a parametric preprocessor enhanced the simulated crashworthiness by around 20%. In a third design experiment, the design space was increased and an additional 7% improvement in simulated crashworthiness was found.The designs were generated several times faster and were less expensive to evaluate than with previous manual methods. Also, by using distributed computing to explore the design space on a network of workstations, the number of designs that could be evaluated was dramatically increased.