This paper describes a pilot study using a prototype telerehabilitation system (Ghostman). Ghostman is a visual augmentation system designed to allow a physical therapist and patient to inhabit each other's viewpoint in an augmented real-world environment. This allows the therapist to deliver instruction remotely and observe performance of a motor skill through the patient's point of view. In a pilot study, we investigated the efficacy of Ghostman by using it to teach participants to use chopsticks. Participants were randomized to a single training session, receiving either Ghostman or face-to-face instructions by the same skilled instructor. Learning was assessed by measuring retention of skills at 24-hour and 7-day post instruction. As hypothesised, there were no differences in reduction of error or time to completion between participants using Ghostman compared to those receiving face-to-face instruction. These initial results in a healthy population are promising and demonstrate the potential application of this technology to patients requiring learning or relearning of motor skills as may be required following a stroke or brain injury.