Impairments in walking speed, capacity, and endurance are commonly seen after stroke. Treadmill training improves endurance and gait speed. However, the lack of variable training speed and automated speed progression increases the risk of backward displacement and falling. An automated, speed-sensing treadmill prototype with partial body weight support, the Variable Automated Speed and Sensing Treadmill II (VASST II), was tested in an outpatient rehabilitation setting. Eleven subacute or chronic hemiplegics who could ambulate at > 0.2 m/s for >50 m participated in the study. All subjects underwent physiotherapist-supervised training on VASST II for 60 min daily, 3 times per week, for 5 weeks (total 15 h). Outcome measures at Week 3 (mid-VASST II training), Week 6 (post-VASST II training), Week 12 (first follow-up), and Week 24 (second follow-up) included the 6 minute walk test (6 MWT), 10 meter walk test (10 MWT), Berg Balance Scale (BBS) score, and Functional Ambulation category (FAC) score. User acceptability of VASST II for both study subjects and physiotherapists were also assessed. All subjects [median (IQR) age: 53.0 (22) years; median (IQR) duration post-stroke: 524 (811) days] completed VASST II training. At baseline, mean ± SD 6 MWT was 114 ± 50.9 m; mean ± SD 10 MWT was 0.37 ± 0.18 m/s; mean ± SD BBS score was 40 ± 10; and, mean ± SD FAC score was 4 ± 1. At Week 6, there were significant improvements in the 6 MWT [158.91 ± 88.69 m; P = 0.003], 10 MWT [0.49 ± 0.30 m/s; P = 0.016], and BBS score [42 ± 10; P = 0.003]. Improvements in 6 MWT and BBS scores were sustained at Week 24, but not in the 10 MWT. No VASST II-training related falls were reported. All subjects rated their VASST II training positively and indicated that it improved their current walking ability. VASST II training was effective, feasible, and safe in patients with subacute or chronic post-stroke hemiparetic gait, with sustained gains in distance walked (6 MWT) and functional balance (BBS score) up to 19 weeks post-intervention.