Testing for vehicle rollover performance has proven to be a formidable challenge for the impact test engineer. Realism and repeatability have been difficult to define, let alone produce. Field accident data do not provide a “typical” rollover situation to imitate. Vehicle kinematics, vehicle damage, and occupant injuries cover a very wide spectrum. This paper will cover the history of rollover testing, the current rollover picture from field accidents, and GM's recent experience with a new rollover test method.