Electric power steering (EPS) systems utilize an electric motor drive (EMD), consisting of an electric motor operated in torque control mode, to provide assistance to the driver in steering the vehicle. The torque control behavior of the EMD, which can vary widely depending on the control architecture employed, the controller tuning and on any uncertainty (or error) in estimation of motor parameters, can thus significantly affect the overall stability and performance of EPS systems. This paper presents a detailed examination of the impact of errors in the estimation of various parameters of the EMD system on the performance and stability of EPS systems, considering two different torque control architectures, namely feedback control which utilizes current measurement, and feedforward control which employs an inverse mathematical model of the motor. Both these control architectures use estimates of machine parameters, including the motor voltage (torque) constant, inductance, and resistance, in addition to the inherent tunable control parameters, for accurately generating the electromagnetic torque command requested by EPS control algorithms. Incorrect estimation of these parameters and the selection of controller tuning parameters can result in the degradation of the stability and performance, both steady state as well as dynamic, of the overall EPS control system. Analytical expressions of the open-loop transfer function of the EPS system considering integrated electromechanical dynamics are presented in this paper and are used to study the sensitivity of the EPS system to variations in operating conditions, controller architectures and tuning. The results illustrate that accurate estimation of machine parameters and proper tuning of the motor torque controller is critical to achieving desirable EPS system performance.