This paper addresses the problem of the on-line adaptation of control parameters, dedicated to a path tracking problem in off-road conditions. Two approaches are proposed to modify the tuning gain of a previously developed adaptive and predictive control law. The first approach is a deterministic method based on the dynamic equations of the system, allowing the adaptation of the settling distance with respect to the robot capabilities depending on the grip conditions and velocity. The second strategy uses a neural network trained with a Covariance Matrix Adaptation Evolution Strategy (CMA-ES) algorithm, in order to optimize the robot’s behavior with respect to an objective function. Each approach uses as input dynamic parameters, estimated from sliding angles and cornering stiffness observers. Both methods are described and compared to results obtained when using constant parameters in order to identify their respective strengths and weaknesses. They have been implemented and tested in real conditions on an off-road mobile robot with varying terrain and trajectories. An in depth analysis of the proposed methods is done, and further insights are obtained in the context of gain tuning for steering controllers in dynamic environments. The performance and transferability of these methods are demonstrated, as well as their robustness to changes in the terrain properties. As a result, tracking errors are reduced while preserving the stability and the explainability of the control architecture.