We present the results of a feasibility study for the use of a commercially available non-immersive virtual reality system (SeeMe®) in upper limb motor rehabilitation of stroke survivors. The study included 8 chronic phase stroke survivors. All patients received 12 sessions with the system over 2 weeks, each session comprising 20 minutes of motor training using a sequence of 8 serious games. Outcome measures included the Fugl-Meyer assessment scale -upper limb section (FMA), the Stroke Impact Scale (SIS), the ingame assessment procedure of the SeeMe system and a feedback questionnaire. Following the rehabilitation program, we found no change on the clinical scales. Although some of the system generated parameters (score, endurance and reaction time or movement amplitude for each game) have improved significantly, in 3 of the 8 games none of the recorded variables has shown significant changes, with only one game showing significant improvement in 2 out of 3 parameters. The satisfaction questionnaire did not generally correlate with game performance (although this has happened in some of the games), but reflected correctly the increased interest of the subjects for the intervention and also their awareness of its real influence on motor abilities. This initial pilot study indicates that the SeeMe virtual reality system has the potential to be of use in clinical settings as a complement to conventional therapy. Future studies should include larger number of subjects, longer training duration, use more sensitive/dedicated measurements of improvement, and focus on a single game/exercise type.