The purpose of this study was to investigate the effects of virtual reality (VR) therapy with compensation inhibition and feedback (CIF) on upper extremity function in chronic stroke patients. Seven chronic stroke patients participated in this study, which was a randomized controlled trial with a crossover design. Self upper extremity exercise, conservative VR therapy, and VR therapy with CIF were performed for one hour per session, 5 times per week, over a 3 week period. The main outcome measures involved range of motion (ROM) including shoulder, elbow, and wrist joints, a Manual Function Test (MFT), and a Motor Activity Log (MAL). Data were calculated as posttest and pretest changes in every session and were analyzed using Friedman and Wilcoxon signed-rank tests at p<.05. The results were as follows: 1) Statistically significant increase in ROM measurements of shoulder and elbow joints were seen with VR therapy with CIF compared to VR therapy and self upper extremity exercise (p<.05), whereas no significant increasing was noted for the wrist joint (p>.05). 2) Statistically significant increase in the MFT was seen with VR therapy with CIF compared with VR therapy and self upper extremity exercise (p<.05). 3) VR therapy with CIF also resulted in statistically significant increase in both activity of use (AOU) (p<.05) and quality of movement (QOM) (p<.05) on the MAL test when compared with VR therapy and self upper extremity exercise, respectively. In conclusion, VR therapy with CIF was more effective than conservative VR therapy and self upper extremity exercise in improving the upper extremity function in hemiplegic patients with chronic stroke.