Background: Despite arm rehabilitation often being a neurorehabilitation focus, little is known about unilateral and bilateral arm activity patterns of patients undergoing rehabilitation in hospital settings. Aim: To examine differences in arm activity patterns in rehabilitation inpatients with affected (stroke patients) and unaffected (lower limb orthopaedic patients with no known arm impairment) arm function. Methods: Wrist-mounted tri-axial accelerometers recorded continuous accelerometer data from each wrist for four consecutive, complete days (Thursday-Sunday). Data were split into: rehabilitation clinician working hours (9am-5pm), evening (6pm-8pm), and overnight (midnight-4am). Differences in magnitude and duration of activity were assessed using Cohen’s d effect size (ES). Results: The stroke cohort (n=40) had significantly (p<0.05), lower overall arm activity magnitude compared to the orthopaedic cohort (n=15) of large effect size: affected arm (ES=1.11,-53%),unaffected arm (ES=0.77,-33%) and bilateral (ES=0.99,-42%). Arm activity duration differences were less noticeable. The largest differences between cohorts were observed during the evenings compared to during rehabilitation hours; both magnitude and duration were noticeably higher in the orthopaedic group. The orthopaedic group had predominantly, moderate effect size (mean ES=0.53), significantly lower activity magnitude and duration for both arms used unilaterally and bilaterally on days without compared to with scheduled rehabilitation. Conversely, negligible magnitude (all ES<0.21), non-significant differences for all arm activity magnitude and duration measures existed in the stroke cohort on days without compared to with scheduled rehabilitation. Conclusion: Unexpectedly, arm activity for the stroke cohort was not noticeably lower on days without scheduled rehabilitation. This may have implications for design of rehabilitation programs.