Background and Purpose - To evaluate the effectiveness of a device that automates Constraint-Induced Movement therapy (CI therapy), termed AutoCITE, when only partially supervised by therapists. Methods - Twenty-seven participants with chronic stroke trained with AutoCITE for 3 hours per day for 10 consecutive weekdays. Participants were assigned to 1 of 3 groups in a fixed irregular order (ie, in alternating blocks): supervision from a therapist for 100%, 50%, or 25% of training time. Results - The effect sizes of the treatment gains for the 3 groups on the Motor Activity Log (MAL) were very large and for the Wolf Motor Function Test they were large (all P<0.001) but were not significantly different from one another. Gains were comparable to those previously reported for participants who received an equal amount of standard one-on-one CI therapy without the device. At 1-month and long-term follow-up, gains from pretreatment on the MAL were also significant (P<0.001). Conclusion - These results demonstrate that AutoCITE training with greatly reduced supervision from a therapist is as effective as standard one-on-one CI therapy. © 2005 American Heart Association, Inc.