Motor vehicle collision (MVC) is one of the most common causes of thoracolumbar junction (TLJ) injury. Although it is of no doubt that the use of seatbelt reduces the incidence and severity of MVC-induced TLJ injury, how it is protective for front-seat occupants of an automobile after rollover crashes is unclear. Among 200 consecutive patients with a major TLJ (Th11-L2) injury due to high-energy trauma admitted from 2000 to 2004, 22 patients were identified as front-seat occupants of a four-wheel vehicle when a rollover crash occurred. The 22 patients were divided into two groups: 10 who were belted, and 12 who were unbelted. Patients' demographics including the mean Injury Severity Score (ISS), incidence of neurologic deficit, level of TLJ injury, and type of TLJ injury according to the AO fracture classification were compared between the two groups. Neurologic deficit was present exclusively in the unbelted group, and the difference in the incidence was statistically significant (P = 0.04). Similarly, AO type B/C injury was present exclusively in the unbelted group. The belted group had a significantly lower mean ISS than the unbelted group (P < 0.01). Comparison between the ejected and non-ejected victims within the unbelted group revealed no statistical difference in the incidence of neurologic deficit or type of injury. It is likely that the high incidence of neurologic deficit in the unbelted group was due to the high incidence of AO type B/C injury. This study indirectly proves the efficacy of seatbelt in reducing the severity of rollover-induced TLJ injury. Because of the limited number of cases, it is uncertain whether ejection from vehicle, which occurs exclusively in the unbelted victims, is a crucial factor in determining the severity or type of injury after rollover crashes.