To determine whether use of a robotic arm trainer for bilateral exercise in daily repetitive training for a 3-week period reduced spasticity and improved motor control in the arm of severely affected, chronic hemiparetic subjects. Before-after trial. Community rehabilitation center in Germany. Consecutive sample of 12 chronic hemiparetic patients; minimum stroke interval 6 months; patients could maximally protract the affected shoulder, hold the extended arm, or slightly flex and extend the elbow. Additional daily therapy of 15 minutes with the arm trainer for 3 weeks; the 1 degree of freedom trainer enabled the bilateral passive and active practice of a forearm pronation and supination and wrist dorsiflexion and volarflexion; impedance control guaranteed a smooth movement. Patients' impressions, the Modified Ashworth Scale (MAS) score (range, 0-5) to assess spasticity, and the arm section of the Rivermead Motor Assessment (RMA) score (range, 0-15) to assess motor control were rated before therapy, after each 3-week interval, and at follow-up 3 months later. All patients had favorable impressions: the extremity felt more vivid, and 8 subjects noticed a reduction in spasticity, an ease of hand hygiene, and pain relief. The MAS score of the wrist and fingers joints decreased significantly (P<.0125) from a median of 3 (2-3) and 3 (3-4) to 2 (1-2) and 2.5 (2-3). The RMA score minimally increased in 5 cases without improvement in functional tasks. The median RMA score before therapy was 2.0 (1-2) and 2.0 (1-3.75) after therapy. There were no side effects. At follow-up, the effects had waned. The arm trainer made possible intensive bilateral elbow and wrist training of severely affected stroke patients. Future studies should address the treatment effect in subacute stroke patients and determine the optimum treatment intensity.