As circuit testing days are expensive and limited by regulations, racing teams are more and more dependent on simulation tools. Van Amersfoort Racing built their own racing simulator to train drivers in an fully controlled environment. This environment is based on commercially available simulation software rFactor. However, no research on the accuracy of the physics of this software is available. Since level of fidelity of race simulators is important for the perception of racing drivers, force feedback steering forces are analyzed. Information of the real Formula 3 car is used to upgrade the vehicle model used in rFactor and to develop a Multibody Dynamic vehicle model of the same car. Steering metrics are used to make qualitative comparisons between steering force measurement in the real car, the simulator and the Multibody Dynamic model. It is shown that the baseline simulator vehicle model is less sensitive to steering input compared to the real car. Furthermore the simulator driver theoretically senses higher steering torques for a given lateral acceleration discarding electric power limitations of the force feedback motor. As a desire to improve simulator fidelity, a Pacejka tyre model of the Hankook Formula 3 tyre is converted to an rFactor model together with an improved suspension model using the exact suspension geometry as provided by car manufacturer Dallara. Simultaneously, the Multibody Dynamic vehicle model is constructed from these submodels, which purely focusses on lateral dynamics. In order to use the lateral based Multibody Dynamic model as a tool for simulation and assessment, its response is tested given the same input as the real Formula 3 car experienced during a particular test. Three cases are considered: weaving on a straight, a low speed corner and a high speed corner. Longitudinal load transfer is inherent in low speed corners, which, due to its limitation in the Multibody Dynamic model, leaves the model adjustments inconclusive. Furthermore, tyre relaxation plays an important role in low speed corners following each other up in a short period of time, which affect low speed steering metrics. The Multibody Dynamic model showed close correlation of steering metrics with real car measurements for the high speed corner. The updated rFactor model improved steering torque feedback despite higher required steering angles.