To analyse the effects of gait therapy for patients after acute stroke in a randomized controlled trial. Fifty-six patients with a mean of 8 days post-stroke participated in: (i) gait trainer exercise; (ii) walking training over ground; or (iii) conventional treatment. Patients in the gait trainer exercise and walking groups practiced gait for 15 sessions over 3 weeks and received additional physiotherapy. Functional Ambulatory Category and several secondary outcome measures assessing gait and mobility were administered before and after rehabilitation and at 6-month follow-up. Patients also evaluated their own effort. Walking ability improved more with intensive walk training compared with conventional treatment; median Functional Ambulatory Category was zero in all patients at the start of the study, but it was 3 in both walk-training groups and 0.5 in the conventional treatment group at the end of the therapy. Median Functional Ambulatory Category was 4 in both walk-training groups and 2.5 in conventional treatment group at 6-month follow-up. Mean accomplished walking distance was not different between the gait trainer exercise and over ground walking groups. Borg scale indicated more effort in over ground walking. Secondary outcomes also indicated improvements. Exercise therapy with walking training improved gait function irrespective of the method used, but the time and effort required to achieve the results favour the gait trainer exercise. Early intensive gait training resulted in better walking ability than did conventional treatment.