Robot-assisted gait training (RAGT) is a promising rehabilitation technique that is increasingly used in the clinic to improve walking ability after a neurological disorder. The effectiveness of RAGT might depend on the customization of the robotic therapy, which in most of the cases is done either manually by the clinical practitioner (MT) or by adaptive controllers developed to automatically adjust the assistance (AT). In this contribution we present a comparison of automatic versus manual tuning of RAGT, where we assessed the differences in the adjustment of the therapy for ten participants with neurological disorders (six stroke, four spinal cord injury). The AT approach reached stable assistance levels quicker than the MT approach. Moreover, the AT ensured a good performance for all subtasks of walking with lower assistance levels than the MT. Future clinical trials need to be performed to show whether these apparent advantages result in better clinical outcomes.