Background <p>A proportion of stroke survivors use their paretic arm less than might be expected based on their level of impairment. The resulting underuse of the paretic arm has a negative impact on participation in neurorehabilitation and functional independence. However, non-use remains poorly understood. One possibility is that prioritising the non-paretic arm reflects a habit, despite residual functional capacity in the paretic arm.</p> Methods <p>30 chronic stroke survivors (Mean Fugl Meyer Upper Limb Score: 28.9 ± 11.3) participated in a simplified version of the forced response paradigm, which reliably identifies the presence of a habit. Participants were asked to choose which arm to use to maximise points scored during a reaching task. During half of the trials, the presumed habit of using the non-paretic arm yielded more points, whereas in the other half using the non-paretic arm incurred a loss of points. Participants completed two versions of this task, once with unlimited response time available and once without.</p> Results <p>Participants scored fewer points in the limited response condition compared to the unlimited response conditions. This difference was driven by a selective increase in the use of the non-paretic arm in trials where the paretic arm yielded more points. The results were not mediated by former hand dominance.</p> Conclusions <p>Our results demonstrate that using the non-paretic arm may reflect a habit response that is more readily triggered in demanding (e.g. time-limited) situations. This may explain why successful neurorehabilitation does not always result in a more functionally useful arm. Our results pave the way for targeted interventions such as habit breaking techniques to be included in clinical practise.</p>