A new design methodology based on data mining theory has been proposed and used in the vehicle crashworthiness design. The method allows exploring the big crash simulation dataset to discover the underlying complicated relationships between response and design variables, and derive design rules based on the structural response to make decisions towards the component design. An S-shaped beam is used as an example to demonstrate the performance of this method. A large amount of simulations are conducted and the results form a big dataset. The dataset is then mined to build a decision tree. Based on the decision trees, the interrelationship among the geometric design variables are revealed, and then the design rules are derived to produce the design cases with good energy absorbing capacity. The accuracy of this method is verified by comparing the data mining model prediction and simulation data. The result indicates that the data mining based methodology could overcome the weakness of traditional design method, i.e. lack of capability in information discovery from big datasets.