Hyge sled tests were conducted using a rear-seat sled fixture to evaluate submarining responses (the lap belt of a lap-shoulder belt restraint loads the abdominal region instead of the pelvis). Objectives of these tests included: an evaluation of methods to determine the occurrence of submarining; an investigation into the influence of restraint system parameters, test severity, and type of anthropomorphic test device on submarining response; and an exploration of the mechanics of submarining.This investigation determined that: 1. Slippage of the lap belt off the pelvis due to dynamic loading of the dummy and the resulting kinematics can cause abdominal loading to the dummy in laboratory crash testing. 2. The 5th female dummy submarined more easily than did the Hybrid ill in the test environment. 3. Motion of the pelvis was controlled using a “pelvic stop”, which reduced the submarining tendency for both the 5th female and Hybrid III dummies. 4. Shortening the buckle strap length (in conjunction with a pelvic stop) increased the submarining “threshold”. 5. Modifying the belt restraint by removing the retractor from the lap-belt and adding a retractor to the shoulder belt (with the standard seat) increased the “threshold” severity for submarining. 6. The occurrence of the belt slipping off the pelvis and loading the abdomen was well correlated to rearward rotation of the pelvis. 7. In addition to standard film analysis, detection of the belt slipping off the pelvis is strongly enhanced by use of a submarining indicating pelvis, pelvic in-line accelerometry, and belt force transducers. 8. The development of a method to unequivocally detect the occurrence of submarining, and to predict the severity of any resulting injury would greatly strengthen the analysis of restraint performance.