Steering feel, or steering torque feedback, is an important aspect of a vehicle’s dynamic behaviour and may become more significant in the transition between conventional and automated control. However, there is very little theoretical understanding of its role. The evaluation of steering feel mainly relies on subjective-objective correlation, which turns the subjective ratings of steering feel into objectively measurable metrics. Consequently, this aspect of vehicle development is time consuming, expensive, and probably suboptimal. Therefore, the aim of the research is to improve theoretical understanding of steering feel by measuring, understanding, and modelling a driver’s subjective and objective responses to steering torque feedback. This work builds upon and complements earlier work that investigated the role of vestibular feedback in car driving. A new driver-steering-vehicle model incorporating steering torque feedback is developed for both linear and nonlinear steering dynamics. The underlying hypothesis is that a human driver obtains an internal mental model of the steering and vehicle dynamics, the neuromuscular dynamics, and the sensory systems, which plays a significant role in sensory perception, cognitive control, and neuromuscular action. The effects of the model parameters on the dynamic behaviour of the driver-steering-vehicle system are demonstrated through a comprehensive parameter study. Experiments are devised and performed on a fixed-base driving simulator to identify the unknown parameters of the model so that the driver model is enabled to represent realistic driving behaviours. The objective and subjective experimental data are analysed using rigorous statistical methods to obtain a fundamental understanding of the driver’s steering control behaviour with different steering properties and driving conditions. In general, it is found that with an increase in steering system friction level, the driver’s steering control performance deteriorates, and the subjective evaluation of steering feel is perceived as worse. An identification procedure is initially developed to fit the linear model predictions to measured steering responses in the linear phase of the experiments. The model is found to fit the measured results well under a wide range of conditions, and the identified parameter values are found to be physically plausible. The validity of the identification procedure to find accurate model parameters and the validity of the model structure to describe realistic driver steering control behaviours are checked against experimental and simulation results. The identification and validation procedures are then adjusted to account for challenges of finding parameter values for the nonlinear model. The model structure is found to accurately predict the deterministic component of a driver’ steering control of a vehicle with nonlinear steering system friction. The identified process noise level is found to increase with the increase in steering system friction. However, the possibility that drivers may use an intermittent and threshold-driven control strategy which might explains the identified increasing trend is not examined. A series of simulations is used to investigate the correlations between the model and the driver’s subjective assessment of the vehicle steering quality. It is shown that the model has the potential to explain and predict the driver’s subjective and objective responses to steering torque feedback.