BACKGROUND: Spinal cord injury (SCI) is a severe neurological disorder associated not only with ongoing medical complications but also with a significant loss of mobility and participation. The introduction of robotic technologies to recover lower limb function has been greatly employed in the rehabilitative practice. The aim of this preliminary report were to evaluate the efficacy, the feasibility and the changes in the mobility and in the de-adaptations of a new rehabilitative protocol for EKSO™ a robotic exoskeleton device in subjects with SCI disease with an impairment of lower limbs assessed by gait analysis and clinical outcomes. METHOD: This is a pilot single case experimental A-B (pre-post) design study. Three cognitively intact voluntary participants with SCI and gait disorders were admitted. All subjects were submitted to a training program of robot walking sessions for 45 min daily over 20 sessions. The spatiotemporal parameters at the beginning (T0) and at the end of treatment (T1) were recorded. Other clinical assessments (6 min walking test and Timed Up and Go test) were acquired at T0 and T1. RESULTS: Robot training were feasible and acceptable and all participants completed the training sessions. All subjects showed improvements in gait spatiotemporal indexes (Mean velocity, Cadence, Step length and Step width) and in 6 min Walking Test (T0 versus T1). CONCLUSIONS: Robot training is a feasible form of rehabilitation for people with SCI. Further investigation regarding long term effectiveness of robot training in time is necessary. TRIAL REGISTRATION: ClinicalTrials.gov NCT02065830.