During the past few years, several approaches to spinal-cord repair have been successfully established in animal models. For their use in trials of spinal-cord injury (SCI) in human beings, specific difficulties that affect the success of clinical trials have to be recognised. First, transection of the spinal cord is commonly applied in animal models, whereas contusion, which generally leads to injury in two to three segments, represents the typical injury mechanism in human beings. Second, the quadrupedal organisation of locomotion in animals and the more complex autonomic functions in human beings, challenge translation of animal behaviour into recovery from SCI in people. Third, the extensive damage of motor neurons and roots associated with spinal-cord contusion is not addressed in current translational studies. This damage has direct implications for rehabilitation strategies and functional outcome. Fourth, there is increasing evidence for a degradation of neuronal function below the level of the lesion in chronic complete SCI. The relevance of this degradation for a regeneration-inducing treatment needs to be investigated. Fifth, the prerequisites to enable appropriate reconnection of regenerating tract fibres in a postacute stage have still to be established.