A total of 49 dynamic sled tests were performed with the Hybrid III 10YO to examine issues relating to child belt fit. The goals of these tests were to evaluate ATD response to realistic belt geometries and belt fit, develop methods for accurate, repeatable evaluation of restraint conditions for older children, identify dependent measures that differentiate between good and poor restraint performance, and relate ATD performance to static belt fit with children. The first series of tests examined the effects of lap belt tension, belt configuration, and seating procedure on dynamic responses of the ATD. The second series of tests examined how different designs of booster seat lap belt guides and shoulder belt guides affect performance. In addition, the ATD’s response to different shoulder belt and lap belt geometries was evaluated. With regard to test procedures, use of a lap/shoulder belt with a sliding latchplate produced similar results to using a lap/shoulder belt with fixed anchorages. Use of a production retractor reduced shoulder belt load, as well as head, neck, and chest measures. Reducing lap belt tension to a more realistic 2 lb (rather than 15 lb) did not have a pronounced effect on ATD kinematics with two different booster seats. The UMTRI seating procedure, which produces ATD postures closer to those measured in real children, also prevents the lap belt from being trapped in the gap between the pelvis and the thigh. Use of the UMTRI seating procedure produces more reclined initial postures and more pronounced chin-to-chest contact. A well-designed booster lap belt guide can maintain good belt position dynamically, even with poor lap belt geometry. Shoulder belt guide designs affect ATD kinematics. However, preventing the shoulder belt from coming out of the shoulder belt guide does not necessarily produce better restraint performance, because the belt can still come off of the ATD shoulder during the event, and stiffening booster seats does not necessarily produce better routing of the shoulder belt dynamically. Shoulder belt scores less than 70 mm produce good torso kinematics with the 10YO ATD, but use of HIC as an injury criterion tends to discourage booster seat designs that produce good belt fit on the 10YO ATD. Lap belt angle affects torso kinematics, with shallower lap belt angles leading to submarining and more vertical lap belt angles leading to rollout. Wider spacing of lap belt anchorages leads to submarining, while narrowing spacing leads to rollout. Both upper and lower belt anchorage locations have a strong effect on ATD kinematics. Although good booster designs can mitigate the consequences of poor vehicle lap belt geometry, boosters cannot always overcome poor shoulder belt geometry to keep the belt on the ATD shoulder dynamically, even when they are able to create good static belt fit. This finding suggests that more attention should be focused on the effects of the wide variability in vehicle upper anchorage locations on belt restraint performance for children. Also, because HIC scores are decreased when the torso belt fit is degraded, use of HIC as an injury criterion for booster testing may lead to worse rather than better booster designs.