QUESTION:In adults undergoing rehabilitation after stroke, does 1 hour of additional active repetitive reaching per day prevent or reduce upper limb contracture? DESIGN:Multi-centre, randomised controlled trial with concealed allocation, assessor blinding, and intention-to-treat analysis. PARTICIPANTS:Fifty adults undergoing rehabilitation after stroke who were unable to actively extend the affected wrist past neutral or were unable to flex the affected shoulder to 90 deg. SETTING:Three inpatient rehabilitation units in Australia. INTERVENTION:Both groups received usual upper limb therapy 5 days a week for 5 weeks. In addition, the experimental group received up to 1 hour a day of active, intensive, repetitive upper limb training using the SMART Arm device 5 days a week for 5 weeks. OUTCOME MEASURES:Measures were collected at baseline (Week 0), after intervention (Week 5) and at follow-up (Week 7). The primary outcomes were passive range of wrist extension, elbow extension, and shoulder flexion at Week 5. The secondary outcomes were: the three primary outcomes measured at Week 7; passive range of shoulder external rotation; arm function; and pain at rest, on movement and during sleep measured at Weeks 5 and 7. RESULTS:Following an average of 2310 reaching repetitions, the mean effect at Week 5 on passive range of wrist extension was 1 deg (95% CI -6 to 8), elbow extension -6 deg (95% CI -12 to -1), and shoulder flexion 5 deg (95% CI -8 to 17). There were no statistically significant or clinically important effects of the intervention on any secondary outcomes. CONCLUSION:In adults who are already receiving task-specific motor training for upper limb rehabilitation following stroke, 5 weeks of up to 1 hour of additional daily active repetitive motor training using the SMART Arm device did not prevent or reduce contracture in upper limb muscles. TRIAL REGISTRATION:ACTRN12614001162606.