Robot-mediated physiotherapy provides a new possibility for improving the outcome of rehabilitation of patients who are recovering from stroke. This study is a review of robot-supported upper limb physiotherapy focusing on the shoulder, elbow, and wrist. A literature search was carried out in PubMed, OVID, and EBSCO for clinical trials with robots providing shoulder, elbow, or wrist therapy. Results concerning motor control, spasticity, functional outcome, and the main features of the studies were evaluated. A total of 178 papers were found. On the basis of inclusion/exclusion criteria, 30 studies remained for evaluation. In these trials, a total of 493 patients received robot-aided therapy. The Fugl-Meyer assessment was the most commonly used motor scale, and in 24 of 27 trials, motor function improved significantly. The application of the Modified Ashworth Scale showed that spasticity decreased significantly in nine of 21 trials. Functional scales were only examined in one-third of the studies with significant changes being found in half of them. The intensity and duration of the interventions as well as the elapsed time poststroke were varied. There are several scales, which were used in only a few trials. Unifying the methodology in robotic trials is desirable. Clarification of the acute/subacute/chronic categories, standardizing the application of certain scales for outcome measure in each trial, use of functional scales, and a clearer description of the interventions are recommended.