Purpose : The purpose of this study was to examine the feasibility and efficiency of balance training program through an interactive video game regimen in people with chronic stroke. Methods : Thirty patients with chronic hemiparetic stroke were recruited. Participants were randomly assigned to either a control group (n=15) or an experimental group (n=15). The control group received the general physical therapy including of strengthening and balance exercise five times a week whereas the experimental group received a program of balance exercise with video game play based on virtual reality as well as the same typical physical therapy. The experimental group received 6 sessions for four weeks. Each session was given 5 minutes. An interactive computer game exercise regimen lasted 30 minutes without rest periods. Outcome measures for weight transfer to paretic side, non-paretic side and sit-squat-speed, sit-squat-length, sit-to-standspeed and sit-to-stand-area for the control group (n=15) and experimental group (n=15) before and after treatment were obtained by using the biorescure. Results: Outcomes demonstrated significant improvement in the experimental group compared with the control group in weight transfer to paretic side, non-paretic side and sit-squat-speed, sit-squat-length, sit-to-stand-speed. No significant training effect was showed in sit-to-stand-area between pretraining and post-training. Conclusion : An interactive computer game exercise based on task oriented approach for balance in chronic stroke were feasible. In other words, This regimen resulted in a greater improvement in dynamic balance for people with chronic stroke.