Sensor fault estimation in an automotive application is crucial for the safe execution and availability of software functions controlling the vehicle. Especially, in situations at the grip limits of the tires the safety of the passengers is most critical and a nonlinear tire model is required for accurate predictions. However, the influence of a nonlinear tire model for sensor fault estimation in an automotive application is not sufficiently analyzed in the literature.In this paper, the influence of a nonlinear tire model on sensor fault estimation during highly dynamic cornering maneuvers is demonstrated. A two-stage Extended Kalman Filter (TSEKF) based on a simple state transformation is used for the estimation of additive and multiplicative faults in longitudinal/lateral acceleration and yaw rate. The proposed estimation algorithm was implemented in real-time on a dSPACE MicroAutoBox II and experimental results with multiple measurements of an Audi S8 (D5) are used for validation of the results. The estimation performance during the measurements, using a linear and a nonlinear tire model, is analyzed and the influence of the state transformation on the accuracy of the filter is investigated.Using a simple nonlinear tire model the state estimation performance under nonlinear operating conditions can be improved by up to 79%, thereby reducing false positive detections significantly. Furthermore, the results show that for the considered system without the use of a state transformation, i.e., neglecting the state-fault cross-covariance, degrades the estimation performance by less than 1%.