Steer-by-wire is an advanced steering system which connects the steering wheel with the front wheel by using motors and sensors. Generating a variable road feel to the driver in a steer by wire system is an important task as there is no mechanical connection; it requires the development of suitable feedback algorithm for producing variable torque feedback in the steering wheel. In present work, Neural Network method is proposed for estimating the current to be supplied to the feedback motor using the vehicle dynamic parameters such as vertical displacement of front axle, self-aligning moment and front wheel angle as input parameters. A simulation study has been carried out for verifying the performance of the proposed method using the vehicle dynamic models available in the literature. It was tested for dry, wet and icy road profiles using the typical model parameters. The simulated vehicle dynamic parameters and estimated motor current values for the different road surfaces using neural network are presented. From the simulation results for the sinusoidal road surface and sinusoidal steering angle driver input, it is clear that the neural network based method is able to produce the varying road feel to the driver for the different road conditions.