Background: Stroke incidence rises with age. A stroke can severely affect walking ability, requiring therapy. Robot-assisted walking therapy (ROB) has been advocated as one form of walking rehabilitation in stroke patients. However, its comparative efficacy remains controversial and three-group comparisons are scant. We compared the effects of ROB, walking training therapy without a robot (WTT) and standard treatment therapy (STT) on clinical and mobility outcomes in acute ischemic stroke patients. Methods: Individuals (n = 45, 71 % males, age 64.4y ±6.34), who have recently experienced an ischemic stroke, were randomized to ROB, WTT or STT. Clinical and mobility outcomes were assessed before and after each intervention (3 weeks, 5 sessions/week) and after 5 weeks of no-intervention follow-up. Results: Outcomes did not differ between groups at baseline (p > 0.05). Modified Rankin Scale (primary outcome), improved (p < 0.05) after ROB and WTT vs. STT. These improvements were retained relative to baseline (p < 0.05) after follow-up. Barthel index, Berg Balance Scale, 10-m walking speed, the distance while walking with and without the robot for six minutes, and center pressure velocity in standing improved most after ROB (all p < 0.001), exceeding the changes after WTT which in turn were greater than the changes after STT (p ≤ 0.040). Conclusion: Older adults shortly after an ischemic stroke can quickly learn to walk with a soft robot and retain substantial clinical and mobility improvements at follow-up.