Submarining is a common cause of injury in severe frontal crashes. Furthermore, submarining protection may limit lap-belt designs, in that it restricts the range of usable lap-belt angles. Therefore a method to assess how close the occupant substitute is to submarining in a crash test is desirable.In this paper the hypothesis was that the maximum value of the angle, in the x-z plane, between the lap-belt and the pelvis during the time-period when the belt force exceeds 3 kN is a measure of the lap-belt slipping off the pelvis in a frontal impact. In order to obtain this angle with high precision, five mechanical sled-tests with the Hybrid-III dummy was transferred into MADYMO. In MADYMO belt slip over the dummy pelvis was modeled by means of two orthogonal, translational joints (i. e. joints with one-dimensional translation as their only degree of freedom) on the right and left Anterior Superior Iliac Spine (A.S.I.S., also called iliac crest wing) and thus bilateral and unilateral belt slip could be assessed.The results support the hypothesis. Thus a belt-to-pelvis angle of 11 degrees in the simulations corresponded to observed lap-belt slippage in the sled tests, whereas the lap-belt acted on the pelvic structures throughout those tests in which the belt-to-pelvis angle did not exceed eight degrees. Results from subsequent simulations point out belt-slack, velocity change (Δv), and initial knee angle as factors strongly related to the tendency for submarining.The conclusion is that the maximum belt-to-pelvis angle during the time-span when the lap-belt force exceeds 3 kN can be used to estimate the tendency for belt slip onto the abdomen and submarining in mathematical simulations of frontal impacts.