Objective: The use of new technologies in rehabilitation, such as virtual reality and/or computerized gaming exercises, may be useful to enable patients to practice intensively in a motivating way. The objective of the present randomized controlled pilot study was to compare the effect of reach training using a target group specific-designed rehabilitation game to time-matched standardized conventional reach training on arm function after stroke. Materials and Methods: Twenty chronic stroke patients were randomized to either the rehabilitation game group or the conventional training group. Both groups received three arm training sessions of 30 minutes each week, during a period of 6 weeks. Arm (the upper extremity part of Fugl-Meyer [FM] assessment) and hand (the Action Research Arm [ARA] test) functions were tested 1 week before (T0) and 1 week after (T1) training. A follow-up measurement was performed at 1 month after T1 (T2). Results: ARA and FM scores improved significantly within both groups. Post hoc comparisons revealed significant increases in test scores between T0 and T1 and between T0 and T2 for both ARA and FM, but not for changes from T1 to T2. There were no significant differences between both groups for either clinical test. Conclusions: The present randomized controlled pilot study showed that both arm and hand function improved as much after training with a rehabilitation game as after time-matched conventional training.