Study design Prospective quasi-experimental study, pre-and post-design. Objectives Assess safety, feasibility, training characteristics and changes in gait function for persons with spinal cord injury (SCI) using the robotic exoskeletons from Ekso Bionics. Setting Nine European rehabilitation centres. Methods Robotic exoskeleton gait training, three times weekly over 8 weeks. Time upright, time walking and steps in the device (training characteristics) were recorded longitudinally. Gait and neurological function were measured by 10 Metre Walk Test (10 MWT), Timed Up and Go (TUG), Berg Balance Scale (BBS), Walking Index for Spinal Cord Injury (WISCI) II and Lower Extremity Motor Score (LEMS). Results Fifty-two participants completed the training protocol. Median age: 35.8 years (IQR 27.5-52.5), men/women: N = 36/16, neurological level of injury: C1-L2 and severity: AIS A-D (American Spinal Injury Association Impairment Scale). Time since injury (TSI) amp;lt; 1 year, N = 25; amp;gt; 1 year, N = 27. No serious adverse events occurred. Three participants dropped out following ankle swelling (overuse injury). Four participants sustained a Category II pressure ulcer at contact points with the device but completed the study and skin normalized. Training characteristics increased significantly for all subgroups. The number of participants with TSI amp;lt; 1 year and gait function increased from 20 to 56% (P=0.004) and 10MWT, TUG, BBS and LEMS results improved (P amp;lt; 0.05). The number of participants with TSI amp;gt; 1 year and gait function, increased from 41 to 44% and TUG and BBS results improved (P amp;lt; 0.05). Conclusions Exoskeleton training was generally safe and feasible in a heterogeneous sample of persons with SCI. Results indicate potential benefits on gait function and balance.