Current restraint systems represent one of the most effective engineering developments which have prevented and mitigated car occupant injuries worldwide. However, they are fixed systems, optimized in design terms around a single crash condition with a single occupant in one sitting position. Population issues are addressed only partially through use of the 5th percentile female and 95th percentile male dummies again in a single crash condition. Real world crash injury studies of current seat belts indicate five limitations to occupant protection: head contacts with steering wheels for drivers, intrusion, rear loading by unrestrained objects, mispositioning of the seat belt and injuries from the belt itself. A restraint system of fixed characteristics cannot address the variations in weight, sitting position, biomechanical tolerance and crash severity which occur in the crash populations. Intelligent restraint systems have the potential to address these varying demands so that protection could be optimized for a specific person, in a specific sitting position in a specific crash. The techniques for achieving these aims are variable pretensioners, discretionary web locks and load limiters, position sensing, and airbags with variable inflation rates and volumes.