The results of a number of previous studies have demonstrated that seat-belted occupants can undergo significant upward and outward excursion during the airborne phase of vehicular rollover, which may place the occupant at risk for injury during subsequent ground contacts. Furthermore, testing using human volunteers, ATDs, and cadavers has shown that increasing tension in the restraint system prior to a rollover event may be of value for reducing occupant displacement. On this basis, it may be argued that pretensioning the restraint system, utilizing technology developed and installed primarily for improving injury outcome in frontal impacts, may modify restrained occupant injury potential during rollover accidents. However, the capacity of current pretensioner designs to positively impact the motion of a restrained occupant during rollover remains unclear. Moreover, the pretensioner characteristics necessary to limit occupant motion and head excursion without exacerbating injury potential associated with restraint contact have not been established. In the present study, we utilized pretensioner testing and computational analysis to evaluate the capacity of pretensioners for altering and reducing occupant head excursion during the early phase of a steer-induced rollover. This study builds upon existing MADYMO models that have been validated for the prediction of occupant head, neck and torso kinematics during the airborne phase of vehicle rollover (Newberry et al., 2005; Lai et al., 2005), as well as experiments which evaluated the occupant motion during the trip phase of steer-induced rollover (Yamaguchi et al., 2005). Although there are many approaches to pretensioner design, most employ pyrotechnic energy that is released when a crash is sensed to reduce the effective length of the restraint system. The capacity of such a pretensioner to retract webbing is likely a function of the seatbelt loads at the time of deployment. In the current study, we conducted bench tests of a typical pyrotechnic retractor-based pretensioner to evaluate its capacity to retract webbing at various loads. It was found that applying a transient force function, as quantified in bench testing, could simulate the pretensioner's performance. Additional bench tests were conducted to characterize the transient force pulses of several retractor pretensioners. To evaluate the efficacy of these pretensioners in limiting head excursion in the early phase of rollover, we simulated a steer-induced rollover event using MADYMO. This simulation focused on the time period from steer initiation until first roof contact, and was validated by comparison of simulated motions of the ATD to those measured during the rollover tip-up tests conducted by Yamaguchi et al. (2005). Vertical and lateral excursion of the head and torso were quantified to evaluate pretensioner performance. Pretensioners with characteristics similar to those found in the existing fleet were found to only modestly affect occupant kinematics and head excursion during the early phase of vehicle rollover.