In the intelligent airbag system, the correct classification of occupant type is the precondition and plays an important role in controlling the airbag release time and inflation strength during emergent accidents. In the paper, the novel rapid occupant classification system is proposed in which tens of pressure sensors are needed to real-time collect pressure distribution data and then the rough sets theory is combined to extract classification knowledge from data features. Furthermore, Experiments have been done to verify its efficiency and effectiviness.