Virtualized reality games offer highly interactive and engaging user experience and therefore game-based approaches (GBVR) may have significant potential to enhance clinical rehabilitation practice as traditional therapeutic exercises are often repetitive and boring, reducing patient compliance. The aim of this study was to investigate if a rehabilitation training programme using GBVR could simultaneously improve both motor skill (MS) and confidence (CON), as they are both important determinants of daily living and physical and social functioning. The study was performed using a nondominant hand motor deficit model in nonambidextrous healthy young adults, whereby dominant and nondominant arms acted as control and intervention conditions, respectively. GBVR training was performed using a commercially available tennis-based game. CON and MS were assessed by having each subject perform a comparable real-world motor task (RWMT) before and after training. Baseline CON and MS for performing the RWMT were significantly lower for the nondominant hand and improved after GBVR training, whereas there were no changes in the dominant (control) arm. These results demonstrate that by using a GBVR approach to address a MS deficit in a real-world task, improvements in both MS and CON can be facilitated and such approaches may help increase patient compliance.