Feedback is an important element in motor learning during rehabilitation therapy following stroke. The objective of this pilot study was to better understand the effect of position feedback during task-oriented reach training of the upper limb in people with chronic stroke. Five subjects participated in the training for 30 minutes three times a week for 6 weeks. During training, subjects performed reaching movements over a predefined path. When deviation from this path occurred, shoulder and elbow joints received position feedback using restraining forces. We recorded the amount of position feedback used by each subject. During pre- and posttraining assessments, we collected data from clinical scales, isometric strength, and workspace of the arm. All subjects showed improvement on one or several kinematic variables during a circular motion task after training. One subject showed improvement on all clinical scales. Subjects required position feedback between 7.4% and 14.7% of training time. Although augmented feedback use was limited, kinematic outcome measures and movement performance during training increased in all subjects, which was comparable with other studies. Emphasis on movement errors at the moment they occur may possibly stimulate motor learning when movement tasks with sufficiently high levels of difficulty are applied.