Background: Stroke results in significant disability, which can be reduced by physical rehabilitation. High levels of repetition and activity are required in rehabilitation, but patients are typically sedentary. Using clinically relevant and fun computer games may be one way to achieve increased activity in rehabilitation. Methods/design: A single-blind randomized controlled trial will be conducted to evaluate the feasibility, efficacy and safety of novel stroke-specific rehabilitation software. This software uses controller-free client interaction and inertial motion sensors. Elements of feasibility include recruitment into the trial, ongoing participation (adherence and dropout), perceived benefit, enjoyment and ease of use of the games. Efficacy will be determined by measuring activity and using upper-limb tasks as well as measures of balance and mobility. The hypothesis that the intervention group will have increased levels of physical activity within rehabilitation and improved physical outcomes compared with the control group will be tested. Discussion: Results from this study will provide a basis for discussion of feasibility of this interactive video technological solution in an inpatient situation. Differences in activity levels between groups will be the primary measure of efficacy. It will also provide data on measures of upper-limb function, balance and mobility. Trial registration: ACTRN12614000427673. Prospectively registered 17 April 2014.