Vehicle rollovers are the most fatal type of vehicle crashes among all road vehicle accidents. Off-road vehicles with a high centre of gravity are highly susceptible to rollovers, which can be either tripped or untripped, caused by external impacts or during high-speed manoeuvres, respectively. Currently, rollover prediction systems are used to indicate any potential wheel lift-off in real time. Subsequently, rollover prevention systems mitigate the impending wheel lift-off. In a severe near-rollover wheel lift-off scenario, the existing anti-rollover control systems are not effective in reinstating the vehicle to safety, as the designed range of operation is limited only to the near-wheel lift-off region. Even after the wheel lift-off occurs, the vehicle can be brought safely back to the ground with the timely intervention of the anti-rollover controller. This study proposes a novel energy-based anti-rollover controller with proportional gain augmentation using steering wheel input to reinstate a vehicle from a high-speed on-road near-rollover scenario with wheels on one side in the lifted-off condition. Use of an energy-based controller facilitates the least complicated implementation with minimum computation time for a severe scenario where the response time available is very small. A nonlinear underactuated inverted double pendulum on a cart model is used for predicting the dynamics of the vehicle in a near-rollover scenario. The design of the energy-based controller is done using the Lyapunov-based controller design method. An enhancement in the performance is obtained with the incorporation of an additional proportional gain controller. The behaviour of the designed anti-rollover controller is studied using the inverted double pendulum on a cart model. The effectiveness of the anti-rollover controller in a real-life scenario was studied using the co-simulation of a sophisticated four-wheeled pick-up model available in TruckSim® vehicle dynamic simulation software and the anti-rollover control system implemented in MATLAB®/Simulink® environment. © IMechE 2019.