The objectives of the study were to determine the most effective and reliable means of restraining children on an ambulance cot and to develop recommended field procedures for emergency medical service providers. A series of crash tests at 48 km/h were conducted using convertible child restraints, car beds, and harness systems tested with 3-year, infant, and 6-year size dummies. Belt configuration and backrest position were varied. A new cot and fastener system significantly improved restraint performance over older systems previously tested. A two-belt attachment with elevated cot backrest was found to be the method with the least performance variability for securing either a convertible child restraint or a car bed. It was concluded that children who weigh up to 18 kg, fit in a convertible child restraint, and can tolerate a semi-upright seated position can be restrained in a convertible child restraint secured with two belts to an ambulance cot. Infants who must lie flat can be restrained in a car bed modified for two seatbelt paths and secured to a cot. In each case, the cot backrest must be elevated, and the cot and anchor system must be crashworthy. None of the harness configurations tested proved to be satisfactory, but an effective system could be developed by following accepted restraint design principles.