The setup of a run-off-road scenario was based on the current knowledge about critical run-off-road situations and accidents. The scenario was initiated by a visual secondary task. During the task, an added clockwise yaw deviation, intended to create the run-off-road scenario, was presented visually but not by the vehicle dynamics or the lateral acceleration of the simulator’s motion system. Results from two experiments show that the drivers frequently neutralised the yaw deviation because of their lack of full attention to the secondary task, the occasionally rough yaw deviation, or a combination of both. Because of the frequently neutralised yaw deviations, the number of steering-wheel interventions from an implemented system, intended to steer back into the lane in run-off-road situations, became limited in number. The system generated in total 14 steering-wheel interventions, ranging from torque levels of 0.3 to 3.7Nm. During ten of the interventions, the driver counteracted the torque with one hand. Nevertheless, the drivers that had experienced the interventions would like to have a system that could steer back to the lane when approaching the road edge, and accept that it takes control of the steering wheel. Further research on shared steering control is required so that driver responses to interventions does not neutralise the intended safety benefit of the system.