In this talk I will describe a novel virtual environment-based (VE) system developed to enhance motor learning, particularly in subjects with neurological impairments. A central feature of the system is the simultaneous display on the computer screen of the prerecorded arm movements of a virtual "teacher" and of the arm movements made by a patient using the device. Both sets of movements are monitored with an electromagnetic tracking device, and can be displayed on the screen as the limb's endpoint or as the whole arm. During training, subjects are asked to imitate the teacher trajectory. The difference between the patient and teacher trajectories can then be quantified and used to provide augmented feedback about performance to the subjects in order to enhance their learning. The system also provides a way to quantitatively assess the kinematic properties of the patients' movements during the entire course of a treatment regime, and thus can be used as a tool for measuring the effects of different variables on motor learning. Examples of system features, training scenes and results of studies of patients with stroke who have used the system will be presented.