Vehicle accident statistics indicate that fatalities due to rollover of sport utility vehicles (SUVs) are of international concern. Soft suspension, combined with high centre-of-gravity, increase SUV rollover propensity. Control strategies for prevention or mitigation of rollover, based on the current state of the vehicle, exist but most experience a delay in switching, due to response times of the controllers and actuators used to control vehicles dynamics. This paper proposes a real-time non-linear vehicle preview model (VPM) capable of accurately predicting future vehicle states based on current states. The preview model is developed and experimentally validated on a SUV. Results indicate that important vehicle states, including roll angle and lateral acceleration, can be accurately predicted up to more than 0.25 s into the future. Vehicle dynamics controllers can use the previewed states instead of the measured states to eliminate delays and make the strategies proactive instead of reactive.