Introduction: Movement related impairments and limitations in walking are common long-term after stroke. This multi-arm randomized controlled trial explored the impact of training with an electromechanically assisted gait training (EAGT) system, i.e., the Hybrid Assistive Limb® (HAL), when integrated with conventional rehabilitation focused on gait and mobility. Material and Methods: Participants, aged 18–70 years with lower extremity paresis but able to walk with manual support or supervision 1–10 years after stroke, were randomized to (A) HAL-training on a treadmill, combined with conventional rehabilitation interventions (HAL-group), or (B) conventional rehabilitation interventions only (Conventional group), 3 days/week for 6 weeks, or (C) no intervention (Control group). Participants in the Control group were interviewed weekly regarding their scheduled training. Primary outcome was endurance in walking quantified by the 6 Minute Walk Test (6MWT). A rater blinded to treatment allocation performed assessments pre- and post-intervention and at follow-ups at 6 and 12 months. Baseline assessment included the National Institute of Health Stroke Scale (NIHSS) and the Modified Ranking Scale (MRS). Secondary outcomes included the Fugl Meyer Assessment- Lower Extremity, 10 Meter Walk Test, Berg Balance Scale (BBS), Barthel Index (BI) and perceived mobility with the Stroke Impact Scale. Results: A total of 48 participants completed the intervention period. The HAL-group walked twice as far as the Conventional group during the intervention. Post-intervention, both groups exhibited improved 6 MWT results, while the Control group had declined. A significant improvement was only found in the Conventional group and when compared to the Control group (Tukey HSD p = 0.022), and not between the HAL group and Conventional group (Tukey HSD p = 0.258) or the HAL- group and the Control group (Tukey HSD p = 0.447). There was also a significant decline in the Conventional group from post-intervention to 6 months follow up (p = 0.043). The best fitting model to predict outcome included initial balance (BBS), followed by stroke severity (NIHSS), and dependence in activity and participation (BI and MRS). Conclusion: Intensive conventional gait training induced significant improvements long-term after stroke while integrating treadmill based EAGT had no additional value in this study sample. The results may support cost effective evidence-based interventions for gait training long-term after stroke and further development of EAGT. Trial registration: Published on clinicaltrials.gov (NCT02545088) August 24, 2015.