Regaining one's ability to walk is of great importance for neurological patients and is a major goal of all rehabilitation programs. Gait training of severely affected patients after the neurological event is technically difficult because of their motor weakness and balance disturbances. An innovative locomotor training that incorporates high repetitions of task-oriented practice by the use of body weight-supported treadmill training (BWSTT) was developed to overcome these obstacles. To facilitate the delivery of BWSTT, a motorized robotic driven gait orthosis (robotic-assisted gait training-RAGT) was developed. Two types of robotic gait devices were developed, end-effector and exoskeleton devices. Several randomized controlled trials have been published regarding the usage of RAGT in patients after stroke, spinal cord injury (SCI), multiple sclerosis (MS) and other neurological diseases. According to these trials, the usage of RAGT in combination with conventional rehabilitation treatment has some additive beneficial effect on the ambulation abilities mainly in sub-acute stroke and sub-acute SCI patients. No difference was found between the two types of robotic gait devices. No sufficient data regarding an optimal protocol of RAGT is available, however a longer duration and a higher intensity of RAGT seem to have more beneficial effect on the final functional ambulation outcomes. Larger controlled studies are still required to determine the optimal timing and protocol design for the maximal efficacy and long-term outcome of RAGT in neurological patients.