A review of 1990-92 National Accident Sampling System data found that frontal crashes with direct damage involving two-thirds or less of the front-end of the car are at least as common as crashes with direct damage distributed across the front-end. The role of intrusion in these asymmetric or offset crashes in injury causation —especially to the lower-limbs supports— the need for frontal crash testing that goes beyond the well-established full-width-barrier tests. A series of car-to-car, car-to-rigid barrier, and car-to-deformable barrier crash tests support the use of deformable faces in offset tests as a way of providing a reasonable approximation of actual car-to-car crashes. The results support the growing consensus that a minimum of two tests is needed to assess car crashworthiness in frontal crashes: at full-width-barrier test to assess restraint system performance and an offset test into a deformable barrier to assess structural integrity.