The process of motor learning (ML) involves the reorganization of neural networks mediated by training, which enables an improvement in the performance of the trained task. The success of rehabilitation of patients with nerve damage is closely dependent on the potential of the applied training to activate and increase the ML process. The most limiting and refractory motor sequel of rehabilitation in patients victims of Cerebral Vascular Accident (CVA) is the paresis of the upper limb contralateral to the injured hemisphere. Therefore, the search for new training strategies that enhance the ML process has been a constant in the area. The virtual reality training (VR) is a new promise in this sense, because it could offer favorable conditions to ML. Despite the growing use of this type of tool for rehabilitation of the paretic upper limb (PUL) in patients with stroke, there is no consistent evidence about its potential to promote ML. Thus, the objective of this study was to investigate the improvement of short and long term performances after training for upper limbs (UL) in VR, based on Nintendo Wii® games (NW), in individuals with chronic sequelae of stroke. To achieve this goal, the performance of 22 individuals with chronic sequelae of stroke in UL, 17 men and 5 women in an average age of 52.7 ± 10.29 and 5.68 ± 5.31 years post stroke, was recorded in four NW games that require movements of UL with variable amplitude and speed that were compared before and after training. The intrasession acquisition was obtained by comparing the scores of the games before training (BT) and immediately after training (IAT), while the short-term retention was obtained by comparing the evaluation IAT and 48 hours after training (48hAT), and the long-term between IAT evaluation and 7 days after training (7dAT). The training was conducted in a single session, with five repetitions of each of the four games, in which 2nd, 3rd and 4th, the responsible physiotherapist directed through explicit and implicit clues, in order to benefit the motor responses of patients. The score in the games, taken as the primary endpoint, will be analyzed by ANOVA for repeated measures, considering as a factor assessments (BT, IAT, 48hAT and 7dAT), one for each of the games. We believe that the results of this study can contribute with evidences, substantiating the therapeutic use in VR for stroke patients, based on their potential to promote PUL motor performance improvement and its retention.