Far-side lateral impacts were simulated using a Part 572 dummy and human cadavers to compare responses for several belt restraint configurations. Sled tests were conducted having a velocity change of 35 km/hr at a 10 g deceleration level. It was estimated from field data that a 35 km/hr velocity change of the laterally struck vehicle represents about an 80th percentile level for injury-producing lateral collisions.Subjects restrained by a three-point belt system with an outboard anchored diagonal shoulder belt (i.e., positioned over the shoulder opposite the side of impact) rotated out of the shoulder belt and onto the seat. The subject received some lateral restraint due to interaction with the shoulder belt and seatback. The subjects restrained by a three-point belt system with an inboard anchored diagonal shoulder belt (i.e., positioned over the shoulder on the side of impact) remained essentially upright due to shoulder belt interaction with the neck and/or head. Kinematic responses of the Part 572 dummy were generally similar to those of the cadaver subjects.Injuries were found in cadavers restrained by both shoulder belt configurations, but were more extensive to the cervical region for those subjects receiving direct neck and/or head loading from the belt. However, limitations in the cadaver model and test environment, as well as the preliminary nature of the experiments do not permit definitive conclusions on the significance or applicability of the injury data obtained to real accident situations at this time.