In non-disabled individuals, arm choice in pointing movements depends on expected biomechanical effort, expected success, and a handedness bias. Following a stroke, is arm choice re-optimized to account for the decreased motor performance, or does it follow a pre-injury habitual pattern? Because premorbidly right-handed individuals with left hemiparesis generally use their affected arm less than those with right hemiparesis, we hypothesized that arm choice follows a more habitual pattern in right- than in left-hemiparetic individuals. Participants with mild to moderate chronic stroke who were right-handed before stroke performed pointing movements in both free- and forced-choice blocks, both under a no-time constraint condition and under a fast-time constraint condition designed to promote habitual choice. Mixed-effects models of arm choice revealed that expected effort and side of stroke predicted choices overall. However, expected success predicted choice in left-, but not of right-hemiparetic individuals. Furthermore, while left-hemiparetic individuals tended to avoid unsuccessful movements in the fast condition by selecting their non-paretic arm, right-hemiparetic individuals persevered in choosing their more affected arm. In addition, reaction times decreased in left-hemiparetic individuals between the no-time and the fast condition but showed no changes in right-hemiparetic individuals. Finally, arm choice in the no-time condition correlated with a clinical measure of spontaneous arm use for right-, but not for left-hemiparetic individuals. Our results thus show that, in premorbidly right-handed individuals with mild to moderate chronic stroke, arm choice is habitual in right-hemiparetic individuals, but shows a greater degree of optimality by taking account expected success in left-hemiparetic individuals.