A modeling study was conducted to explore the potential for reducing rollover occupant injury via seat belt geometry modifications and the effect such modifications would have on frontal impact protection. MADYMO software was used to model the first roof-to-ground strike of a dolly rollover crash test as well as a frontal impact test using a facet-style human driver occupant in a sport utility vehicle. The objective of this study was to learn whether occupant protection could be improved for rollover without reducing occupant protection in frontal impact. The models were validated using crash test results. Seat belt anchor locations were independently varied in the models to examine the effects of shortening the length or increasing the angle of the lap belt, shortening the torso belt, or lowering the D-ring. Several combinations of the most promising independent anchor relocations were investigated first in near-side and far-side rollover and then in frontal impact. Time histories of head excursion, head impact velocity, and chest acceleration were considered in assessing injury potential. For the occupant size and seating position examined and for the vehicle dynamics used, two anchor relocation combinations were found to potentially have a favorable effect on occupant protection: a.) reducing the lap belt length and b.) reducing the lap belt length and the shoulder belt length. Lowering the D-ring was found to reduce protection of the far-side rollover occupant who has slipped from the torso belt, because the opportunity for him to re-enter the torso belt is diminished. In addition, it was observed in general that if the occupant's head contacts the roof during the rollover, it tends to do so before the head-roof contact patch contacts the ground. Consequently, the key determinant of head-“ground” contact velocity is then the vehicle dynamics, and modifications to the seat belt contribute to protection only by varying the placement of the head-roof contact. It should be noted that for the rollover scenario explored, the primary roof-ground interaction was with the near-side roof rail. The conclusions reached in this study should not be assumed to apply to a rollover scenario in which the far-side roof rail has a hard impact with the ground.