The current generation of framed Child Restraint Systems (CRS), when used by children in cars, are commonly attached to the structure by lap and diagonal automatic seat belts. In order to conform to ECE R44 (the European regulation) these systems are dynamically tested in simulated vehicle frontal impacts using a standardised test seat and anchorage positions, with static lap and diagonal belts. However when used in cars, the value of certain parameters may differ from those in the compliance tests. This study assesses the performance of a sample of current UK CRS in car body shells, mounted on a sled, compared with baseline CRS performances. All tests were based on the standard ECE R44 impact pulse Differences in initial webbing tension and stiffness and belt anchorage geometry are shown to result in changes in performance of the CRS. Comments are made regarding the influence of seat cushions. As a result of this study some modifications to the current test procedure are proposed.