BACKGROUND: A large proportion of patients with spinal cord injury (SCI) regain ambulatory function. However, during the first 3 months most of the patients are not able to walk unsupported. To enable ambulatory training at such an early stage the body weight is partially relieved and the leg movements are assisted by two therapists. A more recent approach is the application of robotic based assistance which allows for longer training duration. From motor learning science and studies including patients with stroke, it is known that training effects depend on the duration of the training. Longer trainings result in a better walking function. The aim of the present study is to evaluate if prolonged robot assisted walking training leads to a better walking outcome in patients with incomplete SCI and whether such training is feasible or has undesirable effects. METHODS/DESIGN: Patients from multiple sites with a subacute incomplete SCI and who are not able to walk independently will be randomized to either standard training (3-5 sessions per week, session duration maximum 25 minutes) or an intensive training (3-5 sessions per week, session duration minimum 50 minutes). After 8 weeks of training and 4 months later the walking ability, the occurrence of adverse events and the perceived rate of exertion as well as the patients' impression of change will be compared between groups. TRIAL REGISTRATION: This study is registered at clinicaltrials.gov, identifier: NCT01147185.