The effects of caregiver burden during the inpatient rehabilitation period have not yet been investigated. The purpose of this study was to evaluate the burden on stroke survivors' caregivers during the inpatient rehabilitation period, and to compare the associations of robot-assisted gait training and conventional therapy with caregiver burden. Our randomized, crossover, prospective study included 63 stroke survivors and their caregivers, who were randomly assigned to one of two groups. The patients in group I received robot-assisted gait training for 2 weeks, followed by conventional therapy for a further 2 weeks. The patients in group II received conventional therapy for 2 weeks followed by robot-assisted gait training for a further 2 weeks. The caregiver burden inventory, beck depression index, and beck hopelessness scale were administered to the caregivers at day 0, on the "switch day" (day 15), and day 30. Before inpatient rehabilitation, 18 (35%) of the caregivers had somewhat elevated scores on the caregiver burden inventory; however, at the end of rehabilitation, 42 (66.6%) of the caregivers were in a high-burden state. The caregiver burden inventory scores differed significantly between baseline and the end of rehabilitation in both groups. Caregiver depression scores also increased significantly in both groups (p < 0.0001), while hopelessness scores increased only in group II (p = 0.027). Caregiver burden increased during the inpatient stroke rehabilitation period. During inpatient rehabilitation, both robot-assisted gait training and conventional therapy increased caregiver burden. ClinicalTrials.gov Number NCT03535467, First Posted: 24 May 2018.