In this paper, a multi-model observer dedicated to ATV (All-Terrain Vehicle) rollover prevention is proposed. Beyond the stability estimation thanks to an observer, the main contribution of this paper lies in the independent reconstruction of front and rear cornering stiffnesses to describe the vehicle dynamics, with a low cost sensing equipment. This advance allows to take into account the two main vehicle behaviours (over/under-steering) encountered during aggressive manoeuvres in off-road context. Dynamic instability evaluation is then based on the on-line estimation and prediction of the Lateral Load Transfer (LLT) from a vehicle 2D multi-model. More precisely, a predictive control algorithm, relying on the extrapolation of rider’s action and the estimated data from the observer, allows the rollover risk to be anticipated, enabling to warn the pilot and to consider the implementation of active actions. Full-scale experimentations are presented to discuss the efficiency of the proposed solution.