Dynamic response prediction of vehicle bodies is important for vehicle crashworthiness evaluation. The dynamic behaviour of vehicle body materials is dependent on material strain rates. One of the typical high strain rate tensile tests is the split Hopkinson bar test. In this paper, experiments have been conducted based on a new split Hopkinson bar apparatus specially designed for the dynamic tensile test of sheet metals. Results from both quasistatic and dynamic tests show that the strain rate hardening effect for sheet metals cannot be described by the original Johnson-Cook constitutive relation. This relation has been modified to include a higher-order term for the hardening effect. The modified constitutive relation represents a more accurate simulation than the original model for the dynamic behaviour of vehicle body structures.